chore: bump version to 0.2.46 [skip ci]

The method is cfg(unix) but the call site wasn't gated, causing a
compile error on Windows.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

.gitea/workflows/build-sidecar.yml

@@ -0,0 +1,402 @@
+name: Build Sidecars
+
+on:
+  push:
+    branches: [main]
+    paths: ['python/**']
+  workflow_dispatch:
+
+jobs:
+  bump-sidecar-version:
+    name: Bump sidecar version and tag
+    if: "!contains(github.event.head_commit.message, '[skip ci]')"
+    runs-on: ubuntu-latest
+    outputs:
+      version: ${{ steps.bump.outputs.version }}
+      tag: ${{ steps.bump.outputs.tag }}
+      has_changes: ${{ steps.check_changes.outputs.has_changes }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: 2
+
+      - name: Check for python changes
+        id: check_changes
+        run: |
+          # If triggered by workflow_dispatch, always build
+          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            echo "has_changes=true" >> $GITHUB_OUTPUT
+            exit 0
+          fi
+          # Check if any python/ files changed in this commit
+          CHANGED=$(git diff --name-only HEAD~1 HEAD -- python/ 2>/dev/null || echo "")
+          if [ -n "$CHANGED" ]; then
+            echo "has_changes=true" >> $GITHUB_OUTPUT
+            echo "Python changes detected: $CHANGED"
+          else
+            echo "has_changes=false" >> $GITHUB_OUTPUT
+            echo "No python/ changes detected, skipping sidecar build"
+          fi
+
+      - name: Configure git
+        if: steps.check_changes.outputs.has_changes == 'true'
+        run: |
+          git config user.name "Gitea Actions"
+          git config user.email "actions@gitea.local"
+
+      - name: Bump sidecar patch version
+        if: steps.check_changes.outputs.has_changes == 'true'
+        id: bump
+        run: |
+          # Read current version from python/pyproject.toml
+          CURRENT=$(grep '^version = ' python/pyproject.toml | head -1 | sed 's/version = "\(.*\)"/\1/')
+          echo "Current sidecar version: ${CURRENT}"
+
+          # Increment patch number
+          MAJOR=$(echo "${CURRENT}" | cut -d. -f1)
+          MINOR=$(echo "${CURRENT}" | cut -d. -f2)
+          PATCH=$(echo "${CURRENT}" | cut -d. -f3)
+          NEW_PATCH=$((PATCH + 1))
+          NEW_VERSION="${MAJOR}.${MINOR}.${NEW_PATCH}"
+          echo "New sidecar version: ${NEW_VERSION}"
+
+          # Update python/pyproject.toml
+          sed -i "s/^version = \"${CURRENT}\"/version = \"${NEW_VERSION}\"/" python/pyproject.toml
+
+          echo "version=${NEW_VERSION}" >> $GITHUB_OUTPUT
+          echo "tag=sidecar-v${NEW_VERSION}" >> $GITHUB_OUTPUT
+
+      - name: Commit and tag
+        if: steps.check_changes.outputs.has_changes == 'true'
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          NEW_VERSION="${{ steps.bump.outputs.version }}"
+          TAG="${{ steps.bump.outputs.tag }}"
+          git add python/pyproject.toml
+          git commit -m "chore: bump sidecar version to ${NEW_VERSION} [skip ci]"
+          git tag "${TAG}"
+
+          # Push using token for authentication (rebase in case another workflow pushed first)
+          REMOTE_URL=$(git remote get-url origin | sed "s|://|://gitea-actions:${BUILD_TOKEN}@|")
+          git pull --rebase "${REMOTE_URL}" main || true
+          git push "${REMOTE_URL}" HEAD:main
+          git push "${REMOTE_URL}" "${TAG}"
+
+      - name: Create Gitea release
+        if: steps.check_changes.outputs.has_changes == 'true'
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+          TAG="${{ steps.bump.outputs.tag }}"
+          VERSION="${{ steps.bump.outputs.version }}"
+          RELEASE_NAME="Sidecar v${VERSION}"
+
+          curl -s -X POST \
+            -H "Authorization: token ${BUILD_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d "{\"tag_name\": \"${TAG}\", \"name\": \"${RELEASE_NAME}\", \"body\": \"Automated sidecar build.\", \"draft\": false, \"prerelease\": false}" \
+            "${REPO_API}/releases"
+          echo "Created release: ${RELEASE_NAME}"
+
+  build-sidecar-linux:
+    name: Build Sidecar (Linux)
+    needs: bump-sidecar-version
+    if: needs.bump-sidecar-version.outputs.has_changes == 'true'
+    runs-on: ubuntu-latest
+    env:
+      PYTHON_VERSION: "3.11"
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ needs.bump-sidecar-version.outputs.tag }}
+
+      - name: Install uv
+        run: |
+          if command -v uv &> /dev/null; then
+            echo "uv already installed: $(uv --version)"
+          else
+            curl -LsSf https://astral.sh/uv/install.sh | sh
+            echo "$HOME/.local/bin" >> $GITHUB_PATH
+          fi
+
+      - name: Install ffmpeg
+        run: sudo apt-get update && sudo apt-get install -y ffmpeg
+
+      - name: Set up Python
+        run: uv python install ${{ env.PYTHON_VERSION }}
+
+      - name: Build sidecar (CUDA)
+        working-directory: python
+        run: uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --with-cuda
+
+      - name: Package sidecar (CUDA)
+        run: |
+          cd python/dist/voice-to-notes-sidecar && zip -r ../../../sidecar-linux-x86_64-cuda.zip .
+
+      - name: Build sidecar (CPU)
+        working-directory: python
+        run: |
+          rm -rf dist/voice-to-notes-sidecar
+          uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --cpu-only
+
+      - name: Package sidecar (CPU)
+        run: |
+          cd python/dist/voice-to-notes-sidecar && zip -r ../../../sidecar-linux-x86_64-cpu.zip .
+
+      - name: Upload to sidecar release
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          sudo apt-get install -y jq
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+          TAG="${{ needs.bump-sidecar-version.outputs.tag }}"
+
+          # Find the sidecar release by tag (retry up to 30 times with 10s delay)
+          echo "Waiting for sidecar release ${TAG} to be available..."
+          for i in $(seq 1 30); do
+            RELEASE_JSON=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/tags/${TAG}")
+            RELEASE_ID=$(echo "$RELEASE_JSON" | jq -r '.id // empty')
+
+            if [ -n "${RELEASE_ID}" ] && [ "${RELEASE_ID}" != "null" ]; then
+              echo "Found sidecar release: ${TAG} (ID: ${RELEASE_ID})"
+              break
+            fi
+
+            echo "Attempt ${i}/30: Release not ready yet, retrying in 10s..."
+            sleep 10
+          done
+
+          if [ -z "${RELEASE_ID}" ] || [ "${RELEASE_ID}" = "null" ]; then
+            echo "ERROR: Failed to find sidecar release for tag ${TAG} after 30 attempts."
+            exit 1
+          fi
+
+          for file in sidecar-*.zip; do
+            filename=$(basename "$file")
+            encoded_name=$(echo "$filename" | sed 's/ /%20/g')
+            echo "Uploading ${filename} ($(du -h "$file" | cut -f1))..."
+
+            ASSET_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/${RELEASE_ID}/assets" | jq -r ".[] | select(.name == \"${filename}\") | .id // empty")
+            if [ -n "${ASSET_ID}" ]; then
+              curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" \
+                "${REPO_API}/releases/${RELEASE_ID}/assets/${ASSET_ID}"
+            fi
+
+            HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST \
+              -H "Authorization: token ${BUILD_TOKEN}" \
+              -H "Content-Type: application/octet-stream" \
+              -T "$file" \
+              "${REPO_API}/releases/${RELEASE_ID}/assets?name=${encoded_name}")
+            echo "Upload response: HTTP ${HTTP_CODE}"
+          done
+
+  build-sidecar-windows:
+    name: Build Sidecar (Windows)
+    needs: bump-sidecar-version
+    if: needs.bump-sidecar-version.outputs.has_changes == 'true'
+    runs-on: windows-latest
+    env:
+      PYTHON_VERSION: "3.11"
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ needs.bump-sidecar-version.outputs.tag }}
+
+      - name: Install uv
+        shell: powershell
+        run: |
+          if (Get-Command uv -ErrorAction SilentlyContinue) {
+            Write-Host "uv already installed: $(uv --version)"
+          } else {
+            irm https://astral.sh/uv/install.ps1 | iex
+            echo "$env:USERPROFILE\.local\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
+          }
+
+      - name: Install ffmpeg
+        shell: powershell
+        run: choco install ffmpeg -y
+
+      - name: Set up Python
+        shell: powershell
+        run: uv python install ${{ env.PYTHON_VERSION }}
+
+      - name: Install 7-Zip
+        shell: powershell
+        run: |
+          if (-not (Get-Command 7z -ErrorAction SilentlyContinue)) {
+            choco install 7zip -y
+          }
+
+      - name: Build sidecar (CUDA)
+        shell: powershell
+        working-directory: python
+        run: uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --with-cuda
+
+      - name: Package sidecar (CUDA)
+        shell: powershell
+        run: |
+          7z a -tzip -mx=5 sidecar-windows-x86_64-cuda.zip .\python\dist\voice-to-notes-sidecar\*
+
+      - name: Build sidecar (CPU)
+        shell: powershell
+        working-directory: python
+        run: |
+          Remove-Item -Recurse -Force dist\voice-to-notes-sidecar
+          uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --cpu-only
+
+      - name: Package sidecar (CPU)
+        shell: powershell
+        run: |
+          7z a -tzip -mx=5 sidecar-windows-x86_64-cpu.zip .\python\dist\voice-to-notes-sidecar\*
+
+      - name: Upload to sidecar release
+        shell: powershell
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          $REPO_API = "${{ github.server_url }}/api/v1/repos/${{ github.repository }}"
+          $Headers = @{ "Authorization" = "token $env:BUILD_TOKEN" }
+          $TAG = "${{ needs.bump-sidecar-version.outputs.tag }}"
+
+          # Find the sidecar release by tag (retry up to 30 times with 10s delay)
+          Write-Host "Waiting for sidecar release ${TAG} to be available..."
+          $RELEASE_ID = $null
+
+          for ($i = 1; $i -le 30; $i++) {
+            try {
+              $release = Invoke-RestMethod -Uri "${REPO_API}/releases/tags/${TAG}" -Headers $Headers -ErrorAction Stop
+              $RELEASE_ID = $release.id
+
+              if ($RELEASE_ID) {
+                Write-Host "Found sidecar release: ${TAG} (ID: ${RELEASE_ID})"
+                break
+              }
+            } catch {
+              # Release not ready yet
+            }
+
+            Write-Host "Attempt ${i}/30: Release not ready yet, retrying in 10s..."
+            Start-Sleep -Seconds 10
+          }
+
+          if (-not $RELEASE_ID) {
+            Write-Host "ERROR: Failed to find sidecar release for tag ${TAG} after 30 attempts."
+            exit 1
+          }
+
+          Get-ChildItem -Path . -Filter "sidecar-*.zip" | ForEach-Object {
+            $filename = $_.Name
+            $encodedName = [System.Uri]::EscapeDataString($filename)
+            $size = [math]::Round($_.Length / 1MB, 1)
+            Write-Host "Uploading ${filename} (${size} MB)..."
+
+            try {
+              $assets = Invoke-RestMethod -Uri "${REPO_API}/releases/${RELEASE_ID}/assets" -Headers $Headers
+              $existing = $assets | Where-Object { $_.name -eq $filename }
+              if ($existing) {
+                Invoke-RestMethod -Uri "${REPO_API}/releases/${RELEASE_ID}/assets/$($existing.id)" -Method Delete -Headers $Headers
+              }
+            } catch {}
+
+            $uploadUrl = "${REPO_API}/releases/${RELEASE_ID}/assets?name=${encodedName}"
+            $result = curl.exe --fail --silent --show-error `
+              -X POST `
+              -H "Authorization: token $env:BUILD_TOKEN" `
+              -H "Content-Type: application/octet-stream" `
+              -T "$($_.FullName)" `
+              "$uploadUrl" 2>&1
+            if ($LASTEXITCODE -eq 0) {
+              Write-Host "Upload successful: ${filename}"
+            } else {
+              Write-Host "WARNING: Upload failed for ${filename}: ${result}"
+            }
+          }
+
+  build-sidecar-macos:
+    name: Build Sidecar (macOS)
+    needs: bump-sidecar-version
+    if: needs.bump-sidecar-version.outputs.has_changes == 'true'
+    runs-on: macos-latest
+    env:
+      PYTHON_VERSION: "3.11"
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ needs.bump-sidecar-version.outputs.tag }}
+
+      - name: Install uv
+        run: |
+          if command -v uv &> /dev/null; then
+            echo "uv already installed: $(uv --version)"
+          else
+            curl -LsSf https://astral.sh/uv/install.sh | sh
+            echo "$HOME/.local/bin" >> $GITHUB_PATH
+          fi
+
+      - name: Install ffmpeg
+        run: brew install ffmpeg
+
+      - name: Set up Python
+        run: uv python install ${{ env.PYTHON_VERSION }}
+
+      - name: Build sidecar (CPU)
+        working-directory: python
+        run: uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --cpu-only
+
+      - name: Package sidecar (CPU)
+        run: |
+          cd python/dist/voice-to-notes-sidecar && zip -r ../../../sidecar-macos-aarch64-cpu.zip .
+
+      - name: Upload to sidecar release
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          which jq || brew install jq
+
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+          TAG="${{ needs.bump-sidecar-version.outputs.tag }}"
+
+          # Find the sidecar release by tag (retry up to 30 times with 10s delay)
+          echo "Waiting for sidecar release ${TAG} to be available..."
+          for i in $(seq 1 30); do
+            RELEASE_JSON=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/tags/${TAG}")
+            RELEASE_ID=$(echo "$RELEASE_JSON" | jq -r '.id // empty')
+
+            if [ -n "${RELEASE_ID}" ] && [ "${RELEASE_ID}" != "null" ]; then
+              echo "Found sidecar release: ${TAG} (ID: ${RELEASE_ID})"
+              break
+            fi
+
+            echo "Attempt ${i}/30: Release not ready yet, retrying in 10s..."
+            sleep 10
+          done
+
+          if [ -z "${RELEASE_ID}" ] || [ "${RELEASE_ID}" = "null" ]; then
+            echo "ERROR: Failed to find sidecar release for tag ${TAG} after 30 attempts."
+            exit 1
+          fi
+
+          for file in sidecar-*.zip; do
+            filename=$(basename "$file")
+            encoded_name=$(echo "$filename" | sed 's/ /%20/g')
+            echo "Uploading ${filename} ($(du -h "$file" | cut -f1))..."
+
+            ASSET_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/${RELEASE_ID}/assets" | jq -r ".[] | select(.name == \"${filename}\") | .id // empty")
+            if [ -n "${ASSET_ID}" ]; then
+              curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" \
+                "${REPO_API}/releases/${RELEASE_ID}/assets/${ASSET_ID}"
+            fi
+
+            HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST \
+              -H "Authorization: token ${BUILD_TOKEN}" \
+              -H "Content-Type: application/octet-stream" \
+              -T "$file" \
+              "${REPO_API}/releases/${RELEASE_ID}/assets?name=${encoded_name}")
+            echo "Upload response: HTTP ${HTTP_CODE}"
+          done

.gitea/workflows/cleanup-releases.yml

@@ -0,0 +1,65 @@
+name: Cleanup Old Releases
+
+on:
+  # Run after release and sidecar workflows complete
+  schedule:
+    - cron: '0 6 * * *'  # Daily at 6am UTC
+  workflow_dispatch:
+
+jobs:
+  cleanup:
+    name: Remove old releases
+    runs-on: ubuntu-latest
+    env:
+      KEEP_COUNT: 5
+    steps:
+      - name: Cleanup old app releases
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+
+          # Get all releases, sorted newest first (API default)
+          RELEASES=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+            "${REPO_API}/releases?limit=50")
+
+          # Separate app releases (v*) and sidecar releases (sidecar-v*)
+          APP_IDS=$(echo "$RELEASES" | jq -r '[.[] | select(.tag_name | startswith("v") and (startswith("sidecar") | not)) | .id] | .[]')
+          SIDECAR_IDS=$(echo "$RELEASES" | jq -r '[.[] | select(.tag_name | startswith("sidecar-v")) | .id] | .[]')
+
+          # Delete app releases beyond KEEP_COUNT
+          COUNT=0
+          for ID in $APP_IDS; do
+            COUNT=$((COUNT + 1))
+            if [ $COUNT -le ${{ env.KEEP_COUNT }} ]; then
+              continue
+            fi
+            TAG=$(echo "$RELEASES" | jq -r ".[] | select(.id == $ID) | .tag_name")
+            echo "Deleting app release $ID ($TAG)..."
+            curl -s -o /dev/null -w "HTTP %{http_code}\n" -X DELETE \
+              -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/$ID"
+            # Also delete the tag
+            curl -s -o /dev/null -X DELETE \
+              -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/tags/$TAG"
+          done
+
+          # Delete sidecar releases beyond KEEP_COUNT
+          COUNT=0
+          for ID in $SIDECAR_IDS; do
+            COUNT=$((COUNT + 1))
+            if [ $COUNT -le ${{ env.KEEP_COUNT }} ]; then
+              continue
+            fi
+            TAG=$(echo "$RELEASES" | jq -r ".[] | select(.id == $ID) | .tag_name")
+            echo "Deleting sidecar release $ID ($TAG)..."
+            curl -s -o /dev/null -w "HTTP %{http_code}\n" -X DELETE \
+              -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/$ID"
+            curl -s -o /dev/null -X DELETE \
+              -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/tags/$TAG"
+          done
+
+          echo "Cleanup complete. Kept latest ${{ env.KEEP_COUNT }} of each type."

.gitea/workflows/release.yml

@@ -47,9 +47,6 @@ jobs:
           # Update src-tauri/Cargo.toml (match version = "x.y.z" in [package] section)
           sed -i "s/^version = \"${CURRENT}\"/version = \"${NEW_VERSION}\"/" src-tauri/Cargo.toml
 
-          # Update python/pyproject.toml
-          sed -i "s/^version = \".*\"/version = \"${NEW_VERSION}\"/" python/pyproject.toml
-
           echo "new_version=${NEW_VERSION}" >> $GITHUB_OUTPUT
           echo "tag=v${NEW_VERSION}" >> $GITHUB_OUTPUT
 
@@ -58,12 +55,13 @@ jobs:
           BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
         run: |
           NEW_VERSION="${{ steps.bump.outputs.new_version }}"
-          git add package.json src-tauri/tauri.conf.json src-tauri/Cargo.toml python/pyproject.toml
+          git add package.json src-tauri/tauri.conf.json src-tauri/Cargo.toml
           git commit -m "chore: bump version to ${NEW_VERSION} [skip ci]"
           git tag "v${NEW_VERSION}"
 
-          # Push using token for authentication
+          # Push using token for authentication (rebase in case another workflow pushed first)
           REMOTE_URL=$(git remote get-url origin | sed "s|://|://gitea-actions:${BUILD_TOKEN}@|")
+          git pull --rebase "${REMOTE_URL}" main || true
           git push "${REMOTE_URL}" HEAD:main
           git push "${REMOTE_URL}" "v${NEW_VERSION}"
 
@@ -85,42 +83,16 @@ jobs:
   # ── Platform builds (run after version bump) ──
 
   build-linux:
-    name: Build (Linux)
+    name: Build App (Linux)
     needs: bump-version
     runs-on: ubuntu-latest
     env:
-      PYTHON_VERSION: "3.11"
       NODE_VERSION: "20"
     steps:
       - uses: actions/checkout@v4
         with:
           ref: ${{ needs.bump-version.outputs.tag }}
 
-      # ── Python sidecar ──
-      - name: Install uv
-        run: |
-          if command -v uv &> /dev/null; then
-            echo "uv already installed: $(uv --version)"
-          else
-            curl -LsSf https://astral.sh/uv/install.sh | sh
-            echo "$HOME/.local/bin" >> $GITHUB_PATH
-          fi
-
-      - name: Install ffmpeg
-        run: sudo apt-get update && sudo apt-get install -y ffmpeg
-
-      - name: Set up Python
-        run: uv python install ${{ env.PYTHON_VERSION }}
-
-      - name: Build sidecar
-        working-directory: python
-        run: uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --cpu-only
-
-      - name: Package sidecar for Tauri
-        run: |
-          cd python/dist/voice-to-notes-sidecar && zip -r ../../../src-tauri/sidecar.zip .
-
-      # ── Tauri app ──
       - name: Set up Node.js
         uses: actions/setup-node@v4
         with:
@@ -133,7 +105,8 @@ jobs:
 
       - name: Install system dependencies
         run: |
-          sudo apt-get install -y libgtk-3-dev libwebkit2gtk-4.1-dev libappindicator3-dev librsvg2-dev patchelf xdg-utils
+          sudo apt-get update
+          sudo apt-get install -y libgtk-3-dev libwebkit2gtk-4.1-dev libappindicator3-dev librsvg2-dev patchelf xdg-utils rpm
 
       - name: Install npm dependencies
         run: npm ci
@@ -141,7 +114,6 @@ jobs:
       - name: Build Tauri app
         run: npm run tauri build
 
-      # ── Release ──
       - name: Upload to release
         env:
           BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
@@ -150,7 +122,6 @@ jobs:
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
 
           TAG="${{ needs.bump-version.outputs.tag }}"
-          RELEASE_NAME="Voice to Notes ${TAG}"
           echo "Release tag: ${TAG}"
 
           RELEASE_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
@@ -163,7 +134,7 @@ jobs:
 
           echo "Release ID: ${RELEASE_ID}"
 
-          find src-tauri/target/release/bundle -type f -name "*.deb" | while IFS= read -r file; do
+          find src-tauri/target/release/bundle -type f \( -name "*.deb" -o -name "*.rpm" \) | while IFS= read -r file; do
             filename=$(basename "$file")
             encoded_name=$(echo "$filename" | sed 's/ /%20/g')
             echo "Uploading ${filename} ($(du -h "$file" | cut -f1))..."
@@ -184,47 +155,16 @@ jobs:
           done
 
   build-windows:
-    name: Build (Windows)
+    name: Build App (Windows)
     needs: bump-version
     runs-on: windows-latest
     env:
-      PYTHON_VERSION: "3.11"
       NODE_VERSION: "20"
     steps:
       - uses: actions/checkout@v4
         with:
           ref: ${{ needs.bump-version.outputs.tag }}
 
-      # ── Python sidecar ──
-      - name: Install uv
-        shell: powershell
-        run: |
-          if (Get-Command uv -ErrorAction SilentlyContinue) {
-            Write-Host "uv already installed: $(uv --version)"
-          } else {
-            irm https://astral.sh/uv/install.ps1 | iex
-            echo "$env:USERPROFILE\.local\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append
-          }
-
-      - name: Install ffmpeg
-        shell: powershell
-        run: choco install ffmpeg -y
-
-      - name: Set up Python
-        shell: powershell
-        run: uv python install ${{ env.PYTHON_VERSION }}
-
-      - name: Build sidecar
-        shell: powershell
-        working-directory: python
-        run: uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --cpu-only
-
-      - name: Package sidecar for Tauri
-        shell: powershell
-        run: |
-          Compress-Archive -Path python\dist\voice-to-notes-sidecar\* -DestinationPath src-tauri\sidecar.zip
-
-      # ── Tauri app ──
       - name: Set up Node.js
         uses: actions/setup-node@v4
         with:
@@ -249,7 +189,6 @@ jobs:
         shell: powershell
         run: npm run tauri build
 
-      # ── Release ──
       - name: Upload to release
         shell: powershell
         env:
@@ -259,7 +198,6 @@ jobs:
           $Headers = @{ "Authorization" = "token $env:BUILD_TOKEN" }
 
           $TAG = "${{ needs.bump-version.outputs.tag }}"
-          $RELEASE_NAME = "Voice to Notes ${TAG}"
           Write-Host "Release tag: ${TAG}"
 
           $release = Invoke-RestMethod -Uri "${REPO_API}/releases/tags/${TAG}" -Headers $Headers -ErrorAction Stop
@@ -286,7 +224,7 @@ jobs:
               -X POST `
               -H "Authorization: token $env:BUILD_TOKEN" `
               -H "Content-Type: application/octet-stream" `
-              --data-binary "@$($_.FullName)" `
+              -T "$($_.FullName)" `
               "$uploadUrl" 2>&1
             if ($LASTEXITCODE -eq 0) {
               Write-Host "Upload successful: ${filename}"
@@ -296,42 +234,16 @@ jobs:
           }
 
   build-macos:
-    name: Build (macOS)
+    name: Build App (macOS)
     needs: bump-version
     runs-on: macos-latest
     env:
-      PYTHON_VERSION: "3.11"
       NODE_VERSION: "20"
     steps:
       - uses: actions/checkout@v4
         with:
           ref: ${{ needs.bump-version.outputs.tag }}
 
-      # ── Python sidecar ──
-      - name: Install uv
-        run: |
-          if command -v uv &> /dev/null; then
-            echo "uv already installed: $(uv --version)"
-          else
-            curl -LsSf https://astral.sh/uv/install.sh | sh
-            echo "$HOME/.local/bin" >> $GITHUB_PATH
-          fi
-
-      - name: Install ffmpeg
-        run: brew install ffmpeg
-
-      - name: Set up Python
-        run: uv python install ${{ env.PYTHON_VERSION }}
-
-      - name: Build sidecar
-        working-directory: python
-        run: uv run --python ${{ env.PYTHON_VERSION }} python build_sidecar.py --cpu-only
-
-      - name: Package sidecar for Tauri
-        run: |
-          cd python/dist/voice-to-notes-sidecar && zip -r ../../../src-tauri/sidecar.zip .
-
-      # ── Tauri app ──
       - name: Set up Node.js
         uses: actions/setup-node@v4
         with:
@@ -351,7 +263,6 @@ jobs:
       - name: Build Tauri app
         run: npm run tauri build
 
-      # ── Release ──
       - name: Upload to release
         env:
           BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
@@ -361,7 +272,6 @@ jobs:
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
 
           TAG="${{ needs.bump-version.outputs.tag }}"
-          RELEASE_NAME="Voice to Notes ${TAG}"
           echo "Release tag: ${TAG}"
 
           RELEASE_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \

CONTRIBUTING.md

@@ -0,0 +1,140 @@
+# Contributing to Voice to Notes
+
+Thank you for your interest in contributing! This guide covers how to set up the project for development and submit changes.
+
+## Development Setup
+
+### Prerequisites
+
+- **Node.js 20+** and npm
+- **Rust** (stable toolchain)
+- **Python 3.11+** with [uv](https://docs.astral.sh/uv/) (recommended) or pip
+- **System libraries (Linux only):**
+  ```bash
+  sudo apt install libgtk-3-dev libwebkit2gtk-4.1-dev libappindicator3-dev librsvg2-dev patchelf xdg-utils
+  ```
+
+### Clone and Install
+
+```bash
+git clone https://repo.anhonesthost.net/MacroPad/voice-to-notes.git
+cd voice-to-notes
+
+# Frontend
+npm install
+
+# Python sidecar
+cd python && pip install -e ".[dev]" && cd ..
+```
+
+### Running in Dev Mode
+
+```bash
+npm run tauri:dev
+```
+
+This runs the Svelte dev server + Tauri with hot-reload. The Python sidecar runs from your system Python (no PyInstaller needed in dev mode).
+
+### Building
+
+```bash
+# Build the Python sidecar (frozen binary)
+cd python && python build_sidecar.py --cpu-only && cd ..
+
+# Build the full app
+npm run tauri build
+```
+
+## Project Structure
+
+```
+src/                        # Svelte 5 frontend
+  lib/components/           # Reusable UI components
+  lib/stores/               # Svelte stores (app state)
+  routes/                   # SvelteKit pages
+src-tauri/                  # Rust backend (Tauri v2)
+  src/sidecar/              # Python sidecar lifecycle (download, extract, IPC)
+  src/commands/             # Tauri command handlers
+  src/db/                   # SQLite database layer
+python/                     # Python ML sidecar
+  voice_to_notes/           # Main package
+    services/               # Transcription, diarization, AI, export
+    ipc/                    # JSON-line IPC protocol
+    hardware/               # GPU/CPU detection
+.gitea/workflows/           # CI/CD pipelines
+docs/                       # Documentation
+```
+
+## How It Works
+
+The app has three layers:
+
+1. **Frontend (Svelte)** — UI, audio playback (wavesurfer.js), transcript editing (TipTap)
+2. **Backend (Rust/Tauri)** — Desktop integration, file access, SQLite, sidecar process management
+3. **Sidecar (Python)** — ML inference (faster-whisper, pyannote.audio), AI chat, export
+
+Rust and Python communicate via **JSON-line IPC** over stdin/stdout pipes. Each request has an `id`, `type`, and `payload`. The Python sidecar runs as a child process managed by `SidecarManager` in Rust.
+
+## Conventions
+
+### Rust
+- Follow standard Rust conventions
+- Run `cargo fmt` and `cargo clippy` before committing
+- Tauri commands go in `src-tauri/src/commands/`
+
+### Python
+- Python 3.11+, type hints everywhere
+- Use `ruff` for linting: `ruff check python/`
+- Tests with pytest: `cd python && pytest`
+- IPC messages: JSON-line format with `id`, `type`, `payload` fields
+
+### TypeScript / Svelte
+- Svelte 5 runes (`$state`, `$derived`, `$effect`)
+- Strict TypeScript
+- Components in `src/lib/components/`
+- State in `src/lib/stores/`
+
+### General
+- All timestamps in milliseconds (integer)
+- UUIDs as primary keys in the database
+- Don't bundle API keys or secrets — those are user-configured
+
+## Submitting Changes
+
+1. Fork the repository
+2. Create a feature branch: `git checkout -b my-feature`
+3. Make your changes
+4. Test locally with `npm run tauri:dev`
+5. Run linters: `cargo fmt && cargo clippy`, `ruff check python/`
+6. Commit with a clear message describing the change
+7. Open a Pull Request against `main`
+
+## CI/CD
+
+Pushes to `main` automatically:
+- Bump the app version and create a release (`release.yml`)
+- Build app installers for all platforms
+
+Changes to `python/` also trigger sidecar builds (`build-sidecar.yml`).
+
+## Areas for Contribution
+
+- UI/UX improvements
+- New export formats
+- Additional AI provider integrations
+- Performance optimizations
+- Accessibility improvements
+- Documentation and translations
+- Bug reports and testing on different platforms
+
+## Reporting Issues
+
+Open an issue on the [repository](https://repo.anhonesthost.net/MacroPad/voice-to-notes/issues) with:
+- Steps to reproduce
+- Expected vs actual behavior
+- Platform and version info
+- Sidecar logs (`%LOCALAPPDATA%\com.voicetonotes.app\sidecar.log` on Windows)
+
+## License
+
+By contributing, you agree that your contributions will be licensed under the [MIT License](LICENSE).

README.md

@@ -1,32 +1,55 @@
 # Voice to Notes
 
-A desktop application that transcribes audio/video recordings with speaker identification, producing editable transcriptions with synchronized audio playback.
+A desktop application that transcribes audio and video recordings with speaker identification, synchronized playback, and AI-powered analysis. Export to SRT, WebVTT, ASS captions, plain text, or Markdown.
 
 ## Features
 
-- **Speech-to-Text Transcription** — Accurate transcription via faster-whisper (Whisper models) with word-level timestamps
+- **Speech-to-Text** — Accurate transcription via faster-whisper with word-level timestamps. Supports 99 languages.
-- **Speaker Identification (Diarization)** — Detect and distinguish between speakers using pyannote.audio
+- **Speaker Identification** — Detect and label speakers using pyannote.audio. Rename speakers for clean exports.
-- **Synchronized Playback** — Click any word to seek to that point in the audio (Web Audio API for instant playback)
+- **GPU Acceleration** — CUDA support for NVIDIA GPUs (Windows/Linux). Falls back to CPU automatically.
-- **AI Integration** — Ask questions about your transcript via OpenAI, Anthropic, or any OpenAI-compatible API (LiteLLM proxies, Ollama, vLLM)
+- **Synchronized Playback** — Click any word to seek. Waveform visualization via wavesurfer.js.
-- **Export Formats** — SRT, WebVTT, ASS captions, plain text, and Markdown with speaker labels
+- **AI Chat** — Ask questions about your transcript. Works with Ollama (local), OpenAI, Anthropic, or any OpenAI-compatible API.
-- **Cross-Platform** — Builds for Linux, Windows, and macOS (Apple Silicon)
+- **Export** — SRT, WebVTT, ASS, plain text, Markdown — all with speaker labels.
+- **Cross-Platform** — Linux, Windows, macOS (Apple Silicon).
+
+## Quick Start
+
+1. Download the installer from [Releases](https://repo.anhonesthost.net/MacroPad/voice-to-notes/releases)
+2. On first launch, choose **CPU** or **CUDA** sidecar (the AI engine downloads separately, ~500MB–2GB)
+3. Import an audio/video file and click **Transcribe**
+
+See the full [User Guide](docs/USER_GUIDE.md) for detailed setup and usage instructions.
 
 ## Platform Support
 
-| Platform | Architecture | Status |
+| Platform | Architecture | Installers |
-|----------|-------------|--------|
+|----------|-------------|------------|
-| Linux    | x86_64      | Supported |
+| Linux    | x86_64      | .deb, .rpm |
-| Windows  | x86_64      | Supported |
+| Windows  | x86_64      | .msi, .exe (NSIS) |
-| macOS    | ARM (Apple Silicon) | Supported |
+| macOS    | ARM (Apple Silicon) | .dmg |
+
+## Architecture
+
+The app is split into two independently versioned components:
+
+- **App** (v0.2.x) — Tauri desktop shell with Svelte frontend. Small installer (~50MB).
+- **Sidecar** (v1.x) — Python ML engine (faster-whisper, pyannote.audio). Downloaded on first launch. CPU (~500MB) or CUDA (~2GB) variants.
+
+This separation means app UI updates don't require re-downloading the sidecar, and sidecar updates don't require reinstalling the app.
 
 ## Tech Stack
 
-- **Desktop shell:** Tauri v2 (Rust backend + Svelte 5 / TypeScript frontend)
+| Component | Technology |
-- **ML pipeline:** Python sidecar (faster-whisper, pyannote.audio) — frozen via PyInstaller for distribution
+|-----------|-----------|
-- **Audio playback:** wavesurfer.js with Web Audio API backend
+| Desktop shell | Tauri v2 (Rust + Svelte 5 / TypeScript) |
-- **AI providers:** OpenAI, Anthropic, OpenAI-compatible endpoints (local or remote)
+| Transcription | faster-whisper (CTranslate2) |
-- **Local AI:** Bundled llama-server (llama.cpp)
+| Speaker ID | pyannote.audio 3.1 |
-- **Caption export:** pysubs2
+| Audio UI | wavesurfer.js |
+| Transcript editor | TipTap (ProseMirror) |
+| AI (local) | Ollama (any model) |
+| AI (cloud) | OpenAI, Anthropic, OpenAI-compatible |
+| Caption export | pysubs2 |
+| Database | SQLite (rusqlite) |
 
 ## Development
 
@@ -34,8 +57,8 @@ A desktop application that transcribes audio/video recordings with speaker ident
 
 - Node.js 20+
 - Rust (stable)
-- Python 3.11+ with ML dependencies
+- Python 3.11+ with uv or pip
-- System: `libgtk-3-dev`, `libwebkit2gtk-4.1-dev` (Linux)
+- Linux: `libgtk-3-dev`, `libwebkit2gtk-4.1-dev`, `libappindicator3-dev`, `librsvg2-dev`
 
 ### Getting Started
 
@@ -44,49 +67,63 @@ A desktop application that transcribes audio/video recordings with speaker ident
 npm install
 
 # Install Python sidecar dependencies
-cd python && pip install -e . && cd ..
+cd python && pip install -e ".[dev]" && cd ..
 
 # Run in dev mode (uses system Python for the sidecar)
 npm run tauri:dev
 ```
 
-### Building for Distribution
+### Building
 
 ```bash
-# Build the frozen Python sidecar
+# Build the frozen Python sidecar (CPU-only)
-npm run sidecar:build
+cd python && python build_sidecar.py --cpu-only && cd ..
 
-# Build the Tauri app (requires sidecar in src-tauri/binaries/)
+# Build with CUDA support
+cd python && python build_sidecar.py --with-cuda && cd ..
+
+# Build the Tauri app
 npm run tauri build
 ```
 
 ### CI/CD
 
-Gitea Actions workflows are in `.gitea/workflows/`. The build pipeline:
+Two Gitea Actions workflows in `.gitea/workflows/`:
 
-1. **Build sidecar** — PyInstaller-frozen Python binary per platform (CPU-only PyTorch)
+**`release.yml`** — Triggers on push to main:
-2. **Build Tauri app** — Bundles the sidecar via `externalBin`, produces .deb/.AppImage (Linux), .msi (Windows), .dmg (macOS)
+1. Bumps app version (patch), creates git tag and Gitea release
+2. Builds lightweight app installers for all platforms (no sidecar bundled)
+
+**`build-sidecar.yml`** — Triggers on changes to `python/` or manual dispatch:
+1. Bumps sidecar version, creates `sidecar-v*` tag and release
+2. Builds CPU + CUDA variants for Linux/Windows, CPU for macOS
+3. Uploads as separate release assets
 
 #### Required Secrets
 
-| Secret | Purpose | Required? |
+| Secret | Purpose |
-|--------|---------|-----------|
+|--------|---------|
-| `TAURI_SIGNING_PRIVATE_KEY` | Signs Tauri update bundles | Optional (for auto-updates) |
+| `BUILD_TOKEN` | Gitea API token for creating releases and pushing tags |
-
-No other secrets are needed for building. AI provider API keys and HuggingFace tokens are configured by end users in the app's Settings.
 
 ### Project Structure
 
 ```
-src/                    # Svelte 5 frontend
+src/                        # Svelte 5 frontend
-src-tauri/              # Rust backend (Tauri commands, sidecar manager, SQLite)
+  lib/components/           # UI components (waveform, transcript editor, settings, etc.)
-python/                 # Python sidecar (transcription, diarization, AI)
+  lib/stores/               # Svelte stores (settings, transcript state)
-  voice_to_notes/       # Python package
+  routes/                   # SvelteKit pages
-  build_sidecar.py      # PyInstaller build script
+src-tauri/                  # Rust backend
-  voice_to_notes.spec   # PyInstaller spec
+  src/sidecar/              # Sidecar process manager (download, extract, IPC)
-.gitea/workflows/       # Gitea Actions CI/CD
+  src/commands/             # Tauri command handlers
+  nsis-hooks.nsh            # Windows uninstall cleanup
+python/                     # Python sidecar
+  voice_to_notes/           # Python package (transcription, diarization, AI, export)
+  build_sidecar.py          # PyInstaller build script
+  voice_to_notes.spec       # PyInstaller spec
+.gitea/workflows/           # CI/CD (release.yml, build-sidecar.yml)
+docs/                       # Documentation
 ```
 
 ## License
 
-MIT
+[MIT](LICENSE)

docs/USER_GUIDE.md

@@ -0,0 +1,240 @@
+# Voice to Notes — User Guide
+
+## Getting Started
+
+### Installation
+
+Download the installer for your platform from the [Releases](https://repo.anhonesthost.net/MacroPad/voice-to-notes/releases) page:
+
+- **Windows:** `.msi` or `-setup.exe`
+- **Linux:** `.deb` or `.rpm`
+- **macOS:** `.dmg`
+
+### First-Time Setup
+
+On first launch, Voice to Notes will prompt you to download its AI engine (the "sidecar"):
+
+1. Choose **Standard (CPU)** (~500 MB) or **GPU Accelerated (CUDA)** (~2 GB)
+   - Choose CUDA if you have an NVIDIA GPU for significantly faster transcription
+   - CPU works on all computers
+2. Click **Download & Install** and wait for the download to complete
+3. The app will proceed to the main interface once the sidecar is ready
+
+The sidecar only needs to be downloaded once. Updates are detected automatically on launch.
+
+---
+
+## Basic Workflow
+
+### 1. Import Audio or Video
+
+- Click **Import Audio** or press **Ctrl+O** (Cmd+O on Mac)
+- **Audio formats:** MP3, WAV, FLAC, OGG, M4A, AAC, WMA
+- **Video formats:** MP4, MKV, AVI, MOV, WebM — audio is automatically extracted
+
+> **Note:** Video file import requires [FFmpeg](#installing-ffmpeg) to be installed on your system.
+
+### 2. Transcribe
+
+After importing, click **Transcribe** to start the transcription pipeline:
+
+- **Transcription:** Converts speech to text with word-level timestamps
+- **Speaker Detection:** Identifies different speakers (if configured — see [Speaker Detection](#speaker-detection))
+- A progress bar shows the current stage and percentage
+
+### 3. Review and Edit
+
+- The **waveform** displays at the top — click anywhere to seek
+- The **transcript** shows below with speaker labels and timestamps
+- **Click any word** in the transcript to jump to that point in the audio
+- The current word highlights during playback
+- **Edit text** directly in the transcript — word timings are preserved
+
+### 4. Export
+
+Click **Export** and choose a format:
+
+| Format | Extension | Best For |
+|--------|-----------|----------|
+| SRT | `.srt` | Video subtitles (most compatible) |
+| WebVTT | `.vtt` | Web video players, HTML5 |
+| ASS/SSA | `.ass` | Styled subtitles with speaker colors |
+| Plain Text | `.txt` | Reading, sharing, pasting |
+| Markdown | `.md` | Documentation, notes |
+
+All formats include speaker labels when speaker detection is enabled.
+
+### 5. Save Project
+
+- **Ctrl+S** (Cmd+S) saves the current project as a `.vtn` file
+- This preserves the full transcript, speaker assignments, and edits
+- Reopen later to continue editing or re-export
+
+---
+
+## Playback Controls
+
+| Action | Shortcut |
+|--------|----------|
+| Play / Pause | **Space** |
+| Skip back 5s | **Left Arrow** |
+| Skip forward 5s | **Right Arrow** |
+| Seek to word | Click any word in the transcript |
+| Import audio | **Ctrl+O** / **Cmd+O** |
+| Open settings | **Ctrl+,** / **Cmd+,** |
+
+---
+
+## Speaker Detection
+
+Speaker detection (diarization) identifies who is speaking at each point in the audio. It requires a one-time setup:
+
+### Setup
+
+1. Go to **Settings > Speakers**
+2. Create a free account at [huggingface.co](https://huggingface.co/join)
+3. Accept the license on **all three** model pages:
+   - [pyannote/speaker-diarization-3.1](https://huggingface.co/pyannote/speaker-diarization-3.1)
+   - [pyannote/segmentation-3.0](https://huggingface.co/pyannote/segmentation-3.0)
+   - [pyannote/speaker-diarization-community-1](https://huggingface.co/pyannote/speaker-diarization-community-1)
+4. Create a token at [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens) (read access is sufficient)
+5. Paste the token in Settings and click **Test & Download Model**
+
+### Speaker Options
+
+- **Number of speakers:** Set to auto-detect or specify a fixed number for faster results
+- **Skip speaker detection:** Check this to only transcribe without identifying speakers
+
+### Managing Speakers
+
+After transcription, speakers appear as "Speaker 1", "Speaker 2", etc. in the left sidebar. Double-click a speaker name to rename it — the new name appears throughout the transcript and in exports.
+
+---
+
+## AI Chat
+
+The AI chat panel lets you ask questions about your transcript. The AI sees the full transcript with speaker labels as context.
+
+Example prompts:
+- "Summarize this conversation"
+- "What were the key action items?"
+- "What did Speaker 1 say about the budget?"
+
+### Setting Up Ollama (Local AI)
+
+[Ollama](https://ollama.com) runs AI models locally on your computer — no API keys or internet required.
+
+1. **Install Ollama:**
+   - Download from [ollama.com](https://ollama.com)
+   - Or on Linux: `curl -fsSL https://ollama.com/install.sh | sh`
+
+2. **Pull a model:**
+   ```bash
+   ollama pull llama3.2
+   ```
+   Other good options: `mistral`, `gemma2`, `phi3`
+
+3. **Configure in Voice to Notes:**
+   - Go to **Settings > AI Provider**
+   - Select **Ollama**
+   - URL: `http://localhost:11434` (default, usually no change needed)
+   - Model: `llama3.2` (or whichever model you pulled)
+
+4. **Use:** Open the AI chat panel (right sidebar) and start asking questions
+
+### Cloud AI Providers
+
+If you prefer cloud-based AI:
+
+**OpenAI:**
+- Select **OpenAI** in Settings > AI Provider
+- Enter your API key from [platform.openai.com/api-keys](https://platform.openai.com/api-keys)
+- Default model: `gpt-4o-mini`
+
+**Anthropic:**
+- Select **Anthropic** in Settings > AI Provider
+- Enter your API key from [console.anthropic.com](https://console.anthropic.com)
+- Default model: `claude-sonnet-4-6`
+
+**OpenAI Compatible:**
+- For any provider with an OpenAI-compatible API (vLLM, LiteLLM, etc.)
+- Enter the API base URL, key, and model name
+
+---
+
+## Settings Reference
+
+### Transcription
+
+| Setting | Options | Default |
+|---------|---------|---------|
+| Whisper Model | tiny, base, small, medium, large-v3 | base |
+| Device | CPU, CUDA | CPU |
+| Language | Auto-detect, or specify (en, es, fr, etc.) | Auto-detect |
+
+**Model recommendations:**
+- **tiny/base:** Fast, good for clear audio with one speaker
+- **small:** Best balance of speed and accuracy
+- **medium:** Better accuracy, noticeably slower
+- **large-v3:** Best accuracy, requires 8GB+ VRAM (GPU) or 16GB+ RAM (CPU)
+
+### Debug
+
+- **Enable Developer Tools:** Opens the browser inspector for debugging
+
+---
+
+## Installing FFmpeg
+
+FFmpeg is required for importing video files (MP4, MKV, AVI, etc.). It's used to extract the audio track before transcription.
+
+**Windows:**
+```
+winget install ffmpeg
+```
+Or download from [ffmpeg.org/download.html](https://ffmpeg.org/download.html) and add to your PATH.
+
+**macOS:**
+```
+brew install ffmpeg
+```
+
+**Linux (Debian/Ubuntu):**
+```
+sudo apt install ffmpeg
+```
+
+**Linux (Fedora/RHEL):**
+```
+sudo dnf install ffmpeg
+```
+
+After installing, restart Voice to Notes. FFmpeg is not needed for audio-only files (MP3, WAV, FLAC, etc.).
+
+---
+
+## Troubleshooting
+
+### Video import fails / "FFmpeg not found"
+- Install FFmpeg using the instructions above
+- Make sure `ffmpeg` is in your system PATH
+- Restart Voice to Notes after installing
+
+### Transcription is slow
+- Use a smaller model (tiny or base)
+- If you have an NVIDIA GPU, select CUDA in Settings > Transcription > Device
+- Ensure you downloaded the CUDA sidecar during setup
+
+### Speaker detection not working
+- Verify your HuggingFace token in Settings > Speakers
+- Click "Test & Download Model" to re-download
+- Make sure you accepted the license on all three model pages
+
+### Audio won't play / No waveform
+- Check that the audio file still exists at its original location
+- Try re-importing the file
+- Supported formats: MP3, WAV, FLAC, OGG, M4A, AAC, WMA
+
+### App shows "Setting up Voice to Notes"
+- This is the first-launch sidecar download — it only happens once
+- If it fails, check your internet connection and click Retry

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "voice-to-notes",
-  "version": "0.1.0",
+  "version": "0.2.10",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "voice-to-notes",
-      "version": "0.1.0",
+      "version": "0.2.10",
       "license": "MIT",
       "dependencies": {
         "@tauri-apps/api": "^2",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "voice-to-notes",
-  "version": "0.2.5",
+  "version": "0.2.46",
   "description": "Desktop app for transcribing audio/video with speaker identification",
   "type": "module",
   "scripts": {

python/build_sidecar.py

@@ -119,8 +119,11 @@ def create_venv_and_install(cpu_only: bool) -> Path:
             "--index-url", "https://download.pytorch.org/whl/cpu",
         )
     else:
-        print("[build] Installing PyTorch (default, may include CUDA)")
+        print("[build] Installing PyTorch (CUDA 12.6)")
-        pip_install("torch", "torchaudio")
+        pip_install(
+            "torch", "torchaudio",
+            "--index-url", "https://download.pytorch.org/whl/cu126",
+        )
 
     # Install project and dev deps (includes pyinstaller)
     print("[build] Installing project dependencies")

python/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "voice-to-notes"
-version = "0.2.5"
+version = "1.0.13"
 description = "Python sidecar for Voice to Notes — transcription, diarization, and AI services"
 requires-python = ">=3.11"
 license = "MIT"
@@ -15,6 +15,7 @@ dependencies = [
     "pysubs2>=1.7.0",
     "openai>=1.0.0",
     "anthropic>=0.20.0",
+    "soundfile>=0.12.0",
 ]
 
 [project.optional-dependencies]

python/voice_to_notes.spec

@@ -12,15 +12,17 @@ faster_whisper_datas, faster_whisper_binaries, faster_whisper_hiddenimports = co
     "faster_whisper"
 )
 pyannote_datas, pyannote_binaries, pyannote_hiddenimports = collect_all("pyannote")
+soundfile_datas, soundfile_binaries, soundfile_hiddenimports = collect_all("soundfile")
 
 a = Analysis(
     ["voice_to_notes/main.py"],
     pathex=[],
-    binaries=ctranslate2_binaries + faster_whisper_binaries + pyannote_binaries,
+    binaries=ctranslate2_binaries + faster_whisper_binaries + pyannote_binaries + soundfile_binaries,
-    datas=ctranslate2_datas + faster_whisper_datas + pyannote_datas,
+    datas=ctranslate2_datas + faster_whisper_datas + pyannote_datas + soundfile_datas,
     hiddenimports=[
         "torch",
         "torchaudio",
+        "soundfile",
         "huggingface_hub",
         "pysubs2",
         "openai",
@@ -29,16 +31,21 @@ a = Analysis(
     ]
     + ctranslate2_hiddenimports
     + faster_whisper_hiddenimports
-    + pyannote_hiddenimports,
+    + pyannote_hiddenimports
+    + soundfile_hiddenimports,
     hookspath=[],
     hooksconfig={},
     runtime_hooks=[],
     excludes=[
-        "tkinter", "test", "unittest", "pip", "setuptools",
+        "tkinter", "test", "pip", "setuptools",
         # ctranslate2.converters imports torch at module level and causes
         # circular import crashes under PyInstaller. These modules are only
         # needed for model format conversion, never for inference.
         "ctranslate2.converters",
+        # torchcodec is partially bundled by PyInstaller but non-functional
+        # (missing FFmpeg shared libs). Excluding it forces pyannote.audio
+        # to fall back to torchaudio for audio decoding.
+        "torchcodec",
     ],
     win_no_prefer_redirects=False,
     win_private_assemblies=False,

python/voice_to_notes/hardware/detect.py

@@ -105,14 +105,23 @@ def detect_hardware() -> HardwareInfo:
     # RAM info (cross-platform)
     info.ram_mb = _detect_ram_mb()
 
-    # CUDA detection
+    # CUDA detection — verify runtime libraries actually work, not just torch detection
     try:
         import torch
 
         if torch.cuda.is_available():
-            info.has_cuda = True
+            # Test that CUDA runtime libraries are actually loadable
-            info.cuda_device_name = torch.cuda.get_device_name(0)
+            try:
-            info.vram_mb = torch.cuda.get_device_properties(0).total_mem // (1024 * 1024)
+                torch.zeros(1, device="cuda")
+                info.has_cuda = True
+                info.cuda_device_name = torch.cuda.get_device_name(0)
+                info.vram_mb = torch.cuda.get_device_properties(0).total_mem // (1024 * 1024)
+            except Exception as e:
+                print(
+                    f"[sidecar] CUDA detected but runtime unavailable: {e}. Using CPU.",
+                    file=sys.stderr,
+                    flush=True,
+                )
     except ImportError:
         print("[sidecar] torch not available, GPU detection skipped", file=sys.stderr, flush=True)
 

python/voice_to_notes/ipc/handlers.py

@@ -254,15 +254,15 @@ def make_ai_chat_handler() -> HandlerFunc:
             )
 
         if action == "configure":
-            # Re-create a provider with custom settings
+            # Re-create a provider with custom settings and set it active
             provider_name = payload.get("provider", "")
             config = payload.get("config", {})
             if provider_name == "local":
                 from voice_to_notes.providers.local_provider import LocalProvider
 
                 service.register_provider("local", LocalProvider(
-                    base_url=config.get("base_url", "http://localhost:8080"),
+                    base_url=config.get("base_url", "http://localhost:11434/v1"),
-                    model=config.get("model", "local"),
+                    model=config.get("model", "llama3.2"),
                 ))
             elif provider_name == "openai":
                 from voice_to_notes.providers.openai_provider import OpenAIProvider
@@ -286,6 +286,10 @@ def make_ai_chat_handler() -> HandlerFunc:
                     api_key=config.get("api_key"),
                     api_base=config.get("api_base"),
                 ))
+            # Set the configured provider as active
+            print(f"[sidecar] Configured AI provider: {provider_name} with config: {config}", file=sys.stderr, flush=True)
+            if provider_name in ("local", "openai", "anthropic", "litellm"):
+                service.set_active(provider_name)
             return IPCMessage(
                 id=msg.id,
                 type="ai.configured",

python/voice_to_notes/main.py

@@ -5,6 +5,7 @@ from __future__ import annotations
 import signal
 import sys
 
+
 # CRITICAL: Capture real stdout for IPC *before* importing any ML libraries
 # that might print to stdout and corrupt the JSON-line protocol.
 from voice_to_notes.ipc.protocol import init_ipc

python/voice_to_notes/providers/local_provider.py

@@ -1,4 +1,4 @@
-"""Local AI provider — bundled llama-server (OpenAI-compatible API)."""
+"""Local AI provider — Ollama or any OpenAI-compatible API."""
 
 from __future__ import annotations
 
@@ -9,9 +9,9 @@ from voice_to_notes.providers.base import AIProvider
 
 
 class LocalProvider(AIProvider):
-    """Connects to bundled llama-server via its OpenAI-compatible API."""
+    """Connects to Ollama or any OpenAI-compatible API server."""
 
-    def __init__(self, base_url: str = "http://localhost:8080", model: str = "local") -> None:
+    def __init__(self, base_url: str = "http://localhost:11434/v1", model: str = "llama3.2") -> None:
         self._base_url = base_url.rstrip("/")
         self._model = model
         self._client: Any = None
@@ -24,8 +24,8 @@ class LocalProvider(AIProvider):
             from openai import OpenAI
 
             self._client = OpenAI(
-                base_url=f"{self._base_url}/v1",
+                base_url=self._base_url,
-                api_key="not-needed",  # llama-server doesn't require an API key
+                api_key="ollama",  # Ollama doesn't require a real key
             )
         except ImportError:
             raise RuntimeError(
@@ -47,7 +47,9 @@ class LocalProvider(AIProvider):
         try:
             import urllib.request
 
-            req = urllib.request.Request(f"{self._base_url}/health", method="GET")
+            # Check base URL without /v1 suffix for Ollama root endpoint
+            root_url = self._base_url.replace("/v1", "")
+            req = urllib.request.Request(root_url, method="GET")
             with urllib.request.urlopen(req, timeout=2) as resp:
                 return resp.status == 200
         except Exception:
@@ -55,4 +57,4 @@ class LocalProvider(AIProvider):
 
     @property
     def name(self) -> str:
-        return "Local (llama-server)"
+        return "Ollama"

python/voice_to_notes/services/diarize.py

@@ -20,6 +20,81 @@ from voice_to_notes.utils.ffmpeg import get_ffmpeg_path
 from voice_to_notes.ipc.messages import progress_message
 from voice_to_notes.ipc.protocol import write_message
 
+_patched = False
+
+
+def _patch_pyannote_audio() -> None:
+    """Monkey-patch pyannote.audio.core.io.Audio to use torchaudio.
+
+    pyannote.audio has a bug where AudioDecoder (from torchcodec) is used
+    unconditionally even when torchcodec is not installed, causing NameError.
+    This replaces the Audio.__call__ method with a torchaudio-based version.
+    """
+    global _patched
+    if _patched:
+        return
+    _patched = True
+
+    try:
+        import numpy as np
+        import soundfile as sf
+        import torch
+        from pyannote.audio.core.io import Audio
+
+        # Cache loaded audio to avoid re-reading the entire file for every crop call.
+        # For a 3-hour file, crop is called 1000+ times — without caching, each call
+        # reads ~345MB from disk.
+        _audio_cache: dict[str, tuple] = {}
+
+        def _sf_load(audio_path: str) -> tuple:
+            """Load audio via soundfile with caching."""
+            key = str(audio_path)
+            if key in _audio_cache:
+                return _audio_cache[key]
+            data, sample_rate = sf.read(key, dtype="float32")
+            waveform = torch.from_numpy(np.array(data))
+            if waveform.ndim == 1:
+                waveform = waveform.unsqueeze(0)
+            else:
+                waveform = waveform.T
+            _audio_cache[key] = (waveform, sample_rate)
+            return waveform, sample_rate
+
+        def _soundfile_call(self, file: dict) -> tuple:
+            """Replacement for Audio.__call__."""
+            return _sf_load(file["audio"])
+
+        def _soundfile_crop(self, file: dict, segment, **kwargs) -> tuple:
+            """Replacement for Audio.crop — load file once (cached) then slice.
+
+            Pads short segments with zeros to match the expected duration,
+            which pyannote requires for batched embedding extraction.
+            """
+            duration = kwargs.get("duration", None)
+            waveform, sample_rate = _sf_load(file["audio"])
+            # Convert segment (seconds) to sample indices
+            start_sample = int(segment.start * sample_rate)
+            end_sample = int(segment.end * sample_rate)
+            # Clamp to bounds
+            start_sample = max(0, start_sample)
+            end_sample = min(waveform.shape[-1], end_sample)
+            cropped = waveform[:, start_sample:end_sample]
+            # Pad to expected duration if needed (pyannote batches require uniform size)
+            if duration is not None:
+                expected_samples = int(duration * sample_rate)
+            else:
+                expected_samples = int((segment.end - segment.start) * sample_rate)
+            if cropped.shape[-1] < expected_samples:
+                pad = torch.zeros(cropped.shape[0], expected_samples - cropped.shape[-1])
+                cropped = torch.cat([cropped, pad], dim=-1)
+            return cropped, sample_rate
+
+        Audio.__call__ = _soundfile_call  # type: ignore[assignment]
+        Audio.crop = _soundfile_crop  # type: ignore[assignment]
+        print("[sidecar] Patched pyannote Audio to use soundfile", file=sys.stderr, flush=True)
+    except Exception as e:
+        print(f"[sidecar] Warning: Could not patch pyannote Audio: {e}", file=sys.stderr, flush=True)
+
 
 def _ensure_wav(file_path: str) -> tuple[str, str | None]:
     """Convert audio to 16kHz mono WAV if needed.
@@ -113,6 +188,7 @@ class DiarizeService:
         ]
 
         last_error: Exception | None = None
+        _patch_pyannote_audio()
         for model_name in models:
             try:
                 from pyannote.audio import Pipeline
@@ -212,13 +288,20 @@ class DiarizeService:
         thread.start()
 
         elapsed = 0.0
-        estimated_total = max(audio_duration_sec * 0.5, 30.0) if audio_duration_sec else 120.0
+        estimated_total = max(audio_duration_sec * 0.8, 30.0) if audio_duration_sec else 120.0
-        while not done_event.wait(timeout=2.0):
+        duration_str = ""
-            elapsed += 2.0
+        if audio_duration_sec and audio_duration_sec > 600:
+            mins = int(audio_duration_sec / 60)
+            duration_str = f" ({mins}min audio, this may take a while)"
+        while not done_event.wait(timeout=5.0):
+            elapsed += 5.0
             pct = min(20 + int((elapsed / estimated_total) * 65), 85)
+            elapsed_min = int(elapsed / 60)
+            elapsed_sec = int(elapsed % 60)
+            time_str = f"{elapsed_min}m{elapsed_sec:02d}s" if elapsed_min > 0 else f"{int(elapsed)}s"
             write_message(progress_message(
                 request_id, pct, "diarizing",
-                f"Analyzing speakers ({int(elapsed)}s elapsed)..."))
+                f"Analyzing speakers ({time_str} elapsed){duration_str}"))
 
         thread.join()
 

python/voice_to_notes/services/transcribe.py

@@ -77,11 +77,28 @@ class TranscribeService:
             file=sys.stderr,
             flush=True,
         )
-        self._model = WhisperModel(
+        try:
-            model_name,
+            self._model = WhisperModel(
-            device=device,
+                model_name,
-            compute_type=compute_type,
+                device=device,
-        )
+                compute_type=compute_type,
+            )
+        except Exception as e:
+            if device != "cpu":
+                print(
+                    f"[sidecar] Failed to load on {device}: {e}. Falling back to CPU.",
+                    file=sys.stderr,
+                    flush=True,
+                )
+                device = "cpu"
+                compute_type = "int8"
+                self._model = WhisperModel(
+                    model_name,
+                    device=device,
+                    compute_type=compute_type,
+                )
+            else:
+                raise
         self._current_model_name = model_name
         self._current_device = device
         self._current_compute_type = compute_type
@@ -96,17 +113,22 @@ class TranscribeService:
         compute_type: str = "int8",
         language: str | None = None,
         on_segment: Callable[[SegmentResult, int], None] | None = None,
+        chunk_label: str | None = None,
     ) -> TranscriptionResult:
         """Transcribe an audio file with word-level timestamps.
 
         Sends progress messages via IPC during processing.
+        If chunk_label is set (e.g. "chunk 3/12"), messages are prefixed with it.
         """
-        # Stage: loading model
+        prefix = f"{chunk_label}: " if chunk_label else ""
-        write_message(progress_message(request_id, 0, "loading_model", f"Loading {model_name}..."))
+
+        # Stage: loading model (skip for chunks after the first — model already loaded)
+        if not chunk_label:
+            write_message(progress_message(request_id, 0, "loading_model", f"Loading {model_name}..."))
         model = self._ensure_model(model_name, device, compute_type)
 
         # Stage: transcribing
-        write_message(progress_message(request_id, 10, "transcribing", "Starting transcription..."))
+        write_message(progress_message(request_id, 10, "transcribing", f"{prefix}Starting transcription..."))
 
         start_time = time.time()
         segments_iter, info = model.transcribe(
@@ -159,7 +181,7 @@ class TranscribeService:
                     request_id,
                     progress_pct,
                     "transcribing",
-                    f"Transcribing segment {segment_count} ({progress_pct}% of audio)...",
+                    f"{prefix}Transcribing segment {segment_count} ({progress_pct}% of audio)...",
                 )
             )
 
@@ -254,6 +276,7 @@ class TranscribeService:
                 chunk_result = self.transcribe(
                     request_id, tmp.name, model_name, device,
                     compute_type, language, on_segment=chunk_on_segment,
+                    chunk_label=f"Chunk {chunk_idx + 1}/{num_chunks}",
                 )
 
                 # Offset timestamps and merge

src-tauri/Cargo.lock

@@ -59,6 +59,15 @@ version = "1.0.102"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7f202df86484c868dbad7eaa557ef785d5c66295e41b460ef922eca0723b842c"
 
+[[package]]
+name = "arbitrary"
+version = "1.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c3d036a3c4ab069c7b410a2ce876bd74808d2d0888a82667669f8e783a898bf1"
+dependencies = [
+ "derive_arbitrary",
+]
+
 [[package]]
 name = "async-broadcast"
 version = "0.7.2"
@@ -655,6 +664,17 @@ dependencies = [
  "serde_core",
 ]
 
+[[package]]
+name = "derive_arbitrary"
+version = "1.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1e567bd82dcff979e4b03460c307b3cdc9e96fde3d73bed1496d2bc75d9dd62a"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.117",
+]
+
 [[package]]
 name = "derive_more"
 version = "0.99.20"
@@ -4362,7 +4382,7 @@ checksum = "0b928f33d975fc6ad9f86c8f283853ad26bdd5b10b7f1542aa2fa15e2289105a"
 
 [[package]]
 name = "voice-to-notes"
-version = "0.1.0"
+version = "0.2.2"
 dependencies = [
  "chrono",
  "rusqlite",
@@ -4374,6 +4394,7 @@ dependencies = [
  "tauri-plugin-opener",
  "thiserror 1.0.69",
  "uuid",
+ "zip",
 ]
 
 [[package]]
@@ -5412,12 +5433,41 @@ dependencies = [
  "syn 2.0.117",
 ]
 
+[[package]]
+name = "zip"
+version = "2.4.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fabe6324e908f85a1c52063ce7aa26b68dcb7eb6dbc83a2d148403c9bc3eba50"
+dependencies = [
+ "arbitrary",
+ "crc32fast",
+ "crossbeam-utils",
+ "displaydoc",
+ "flate2",
+ "indexmap 2.13.0",
+ "memchr",
+ "thiserror 2.0.18",
+ "zopfli",
+]
+
 [[package]]
 name = "zmij"
 version = "1.0.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b8848ee67ecc8aedbaf3e4122217aff892639231befc6a1b58d29fff4c2cabaa"
 
+[[package]]
+name = "zopfli"
+version = "0.8.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f05cd8797d63865425ff89b5c4a48804f35ba0ce8d125800027ad6017d2b5249"
+dependencies = [
+ "bumpalo",
+ "crc32fast",
+ "log",
+ "simd-adler32",
+]
+
 [[package]]
 name = "zvariant"
 version = "5.10.0"

src-tauri/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "voice-to-notes"
-version = "0.2.5"
+version = "0.2.46"
 description = "Voice to Notes — desktop transcription with speaker identification"
 authors = ["Voice to Notes Contributors"]
 license = "MIT"
@@ -14,7 +14,7 @@ crate-type = ["staticlib", "cdylib", "rlib"]
 tauri-build = { version = "2", features = [] }
 
 [dependencies]
-tauri = { version = "2", features = ["protocol-asset"] }
+tauri = { version = "2", features = ["protocol-asset", "devtools"] }
 tauri-plugin-opener = "2"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
@@ -24,3 +24,6 @@ zip = { version = "2", default-features = false, features = ["deflate"] }
 thiserror = "1"
 chrono = { version = "0.4", features = ["serde"] }
 tauri-plugin-dialog = "2.6.0"
+reqwest = { version = "0.12", features = ["stream", "json"] }
+futures-util = "0.3"
+bytes = "1"

src-tauri/nsis-hooks.nsh

@@ -0,0 +1,11 @@
+; NSIS uninstall hook for Voice to Notes
+; Removes the sidecar data directory (extracted sidecar binaries + logs)
+; but preserves user data in $PROFILE\.voicetonotes (database, settings, models)
+
+!macro NSIS_HOOK_POSTUNINSTALL
+  ; Remove the Tauri app_local_data_dir which contains:
+  ;   - Extracted sidecar directories (voice-to-notes-sidecar/)
+  ;   - sidecar.log
+  ; Path: %LOCALAPPDATA%\com.voicetonotes.app
+  RMDir /r "$LOCALAPPDATA\com.voicetonotes.app"
+!macroend

src-tauri/src/commands/media.rs

@@ -0,0 +1,152 @@
+use std::path::PathBuf;
+use std::process::Command;
+
+#[cfg(target_os = "windows")]
+use std::os::windows::process::CommandExt;
+
+/// Extract audio from a video file to a WAV file using ffmpeg.
+/// Returns the path to the extracted audio file.
+#[tauri::command]
+pub fn extract_audio(file_path: String, output_path: Option<String>) -> Result<String, String> {
+    let input = PathBuf::from(&file_path);
+    if !input.exists() {
+        return Err(format!("File not found: {}", file_path));
+    }
+
+    // Use provided output path, or fall back to a temp WAV file
+    let stem = input.file_stem().unwrap_or_default().to_string_lossy();
+    let output = match output_path {
+        Some(ref p) => PathBuf::from(p),
+        None => std::env::temp_dir().join(format!("{stem}_audio.wav")),
+    };
+
+    eprintln!(
+        "[media] Extracting audio: {} -> {}",
+        input.display(),
+        output.display()
+    );
+
+    // Find ffmpeg — check sidecar extract dir first, then system PATH
+    let ffmpeg = find_ffmpeg().ok_or("ffmpeg not found. Install ffmpeg or ensure it's in PATH.")?;
+
+    let mut cmd = Command::new(&ffmpeg);
+    cmd.args([
+            "-y",             // Overwrite output
+            "-i",
+            &file_path,
+            "-vn",            // No video
+            "-acodec",
+            "pcm_s16le",      // WAV PCM 16-bit
+            "-ar",
+            "22050",          // 22kHz mono for better playback quality
+            "-ac",
+            "1",              // Mono
+        ])
+        .arg(output.to_str().unwrap())
+        .stdout(std::process::Stdio::null())
+        .stderr(std::process::Stdio::piped());
+
+    // Hide the console window on Windows (CREATE_NO_WINDOW = 0x08000000)
+    #[cfg(target_os = "windows")]
+    cmd.creation_flags(0x08000000);
+
+    let status = match cmd.status() {
+        Ok(s) => s,
+        Err(e) if e.raw_os_error() == Some(13) => {
+            // Permission denied — fix permissions and retry
+            eprintln!("[media] Permission denied on ffmpeg, fixing permissions and retrying...");
+            #[cfg(unix)]
+            {
+                use std::os::unix::fs::PermissionsExt;
+                if let Ok(meta) = std::fs::metadata(&ffmpeg) {
+                    let mut perms = meta.permissions();
+                    perms.set_mode(0o755);
+                    let _ = std::fs::set_permissions(&ffmpeg, perms);
+                }
+                // Also fix ffprobe if it exists
+                let ffprobe = ffmpeg.replace("ffmpeg", "ffprobe");
+                if let Ok(meta) = std::fs::metadata(&ffprobe) {
+                    let mut perms = meta.permissions();
+                    perms.set_mode(0o755);
+                    let _ = std::fs::set_permissions(&ffprobe, perms);
+                }
+            }
+            Command::new(&ffmpeg)
+                .args(["-y", "-i", &file_path, "-vn", "-acodec", "pcm_s16le", "-ar", "22050", "-ac", "1"])
+                .arg(output.to_str().unwrap())
+                .stdout(std::process::Stdio::null())
+                .stderr(std::process::Stdio::piped())
+                .status()
+                .map_err(|e| format!("Failed to run ffmpeg after chmod: {e}"))?
+        }
+        Err(e) => return Err(format!("Failed to run ffmpeg: {e}")),
+    };
+
+    if !status.success() {
+        return Err(format!("ffmpeg exited with status {status}"));
+    }
+
+    if !output.exists() {
+        return Err("ffmpeg completed but output file not found".to_string());
+    }
+
+    eprintln!("[media] Audio extracted successfully");
+    Ok(output.to_string_lossy().to_string())
+}
+
+#[tauri::command]
+pub fn check_file_exists(path: String) -> bool {
+    std::path::Path::new(&path).exists()
+}
+
+#[tauri::command]
+pub fn copy_file(src: String, dst: String) -> Result<(), String> {
+    std::fs::copy(&src, &dst).map_err(|e| format!("Failed to copy file: {e}"))?;
+    Ok(())
+}
+
+#[tauri::command]
+pub fn create_dir(path: String) -> Result<(), String> {
+    std::fs::create_dir_all(&path).map_err(|e| format!("Failed to create directory: {e}"))?;
+    Ok(())
+}
+
+/// Find ffmpeg binary — check sidecar directory first, then system PATH.
+fn find_ffmpeg() -> Option<String> {
+    // Check sidecar extract dir (ffmpeg is bundled with the sidecar)
+    if let Some(data_dir) = crate::sidecar::DATA_DIR.get() {
+        // Read sidecar version to find the right directory
+        let version_file = data_dir.join("sidecar-version.txt");
+        if let Ok(version) = std::fs::read_to_string(&version_file) {
+            let version = version.trim();
+            let sidecar_dir = data_dir.join(format!("sidecar-{version}"));
+            let ffmpeg_name = if cfg!(target_os = "windows") {
+                "ffmpeg.exe"
+            } else {
+                "ffmpeg"
+            };
+            let ffmpeg_path = sidecar_dir.join(ffmpeg_name);
+            if ffmpeg_path.exists() {
+                return Some(ffmpeg_path.to_string_lossy().to_string());
+            }
+        }
+    }
+
+    // Fall back to system PATH
+    let ffmpeg_name = if cfg!(target_os = "windows") {
+        "ffmpeg.exe"
+    } else {
+        "ffmpeg"
+    };
+    if Command::new(ffmpeg_name)
+        .arg("-version")
+        .stdout(std::process::Stdio::null())
+        .stderr(std::process::Stdio::null())
+        .status()
+        .is_ok()
+    {
+        return Some(ffmpeg_name.to_string());
+    }
+
+    None
+}

src-tauri/src/commands/mod.rs

@@ -1,6 +1,8 @@
 pub mod ai;
 pub mod export;
+pub mod media;
 pub mod project;
 pub mod settings;
+pub mod sidecar;
 pub mod system;
 pub mod transcribe;

src-tauri/src/commands/project.rs

@@ -12,7 +12,12 @@ use crate::state::AppState;
 pub struct ProjectFile {
     pub version: u32,
     pub name: String,
-    pub audio_file: String,
+    #[serde(default)]
+    pub audio_file: Option<String>,
+    #[serde(default)]
+    pub source_file: Option<String>,
+    #[serde(default)]
+    pub audio_wav: Option<String>,
     pub created_at: String,
     pub segments: Vec<ProjectFileSegment>,
     pub speakers: Vec<ProjectFileSpeaker>,

src-tauri/src/commands/settings.rs

@@ -32,3 +32,16 @@ pub fn save_settings(settings: Value) -> Result<(), String> {
     fs::write(&path, json).map_err(|e| format!("Cannot write settings: {e}"))?;
     Ok(())
 }
+
+/// Toggle devtools on the main window.
+#[tauri::command]
+pub fn toggle_devtools(app: tauri::AppHandle, open: bool) {
+    use tauri::Manager;
+    if let Some(window) = app.get_webview_window("main") {
+        if open {
+            window.open_devtools();
+        } else {
+            window.close_devtools();
+        }
+    }
+}

src-tauri/src/commands/sidecar.rs

@@ -0,0 +1,258 @@
+use futures_util::StreamExt;
+use serde::Serialize;
+use std::io::Write;
+use tauri::{AppHandle, Emitter};
+
+use crate::sidecar::{SidecarManager, DATA_DIR};
+
+const REPO_API: &str = "https://repo.anhonesthost.net/api/v1/repos/MacroPad/voice-to-notes";
+
+#[derive(Serialize, Clone)]
+struct DownloadProgress {
+    downloaded: u64,
+    total: u64,
+    percent: u8,
+}
+
+#[derive(Serialize)]
+pub struct UpdateInfo {
+    pub current_version: String,
+    pub latest_version: String,
+}
+
+/// Read the locally installed sidecar version from `sidecar-version.txt`.
+/// Returns `None` if the file doesn't exist or can't be read.
+fn read_local_sidecar_version() -> Option<String> {
+    let data_dir = DATA_DIR.get()?;
+    let version_file = data_dir.join("sidecar-version.txt");
+    std::fs::read_to_string(version_file)
+        .ok()
+        .map(|v| v.trim().to_string())
+        .filter(|v| !v.is_empty())
+}
+
+/// Write the sidecar version to `sidecar-version.txt` after a successful download.
+fn write_local_sidecar_version(version: &str) -> Result<(), String> {
+    let data_dir = DATA_DIR.get().ok_or("App data directory not initialized")?;
+    let version_file = data_dir.join("sidecar-version.txt");
+    std::fs::write(&version_file, version)
+        .map_err(|e| format!("Failed to write sidecar version file: {}", e))
+}
+
+/// Fetch releases from the Gitea API and find the latest sidecar release
+/// (one whose tag_name starts with "sidecar-v").
+async fn fetch_latest_sidecar_release(
+    client: &reqwest::Client,
+) -> Result<serde_json::Value, String> {
+    let releases_url = format!("{}/releases?limit=20", REPO_API);
+    let resp = client
+        .get(&releases_url)
+        .header("Accept", "application/json")
+        .send()
+        .await
+        .map_err(|e| format!("Failed to fetch releases: {}", e))?;
+
+    if !resp.status().is_success() {
+        return Err(format!("Failed to fetch releases: HTTP {}", resp.status()));
+    }
+
+    let releases = resp
+        .json::<Vec<serde_json::Value>>()
+        .await
+        .map_err(|e| format!("Failed to parse releases JSON: {}", e))?;
+
+    releases
+        .into_iter()
+        .find(|r| {
+            r["tag_name"]
+                .as_str()
+                .map_or(false, |t| t.starts_with("sidecar-v"))
+        })
+        .ok_or_else(|| "No sidecar release found".to_string())
+}
+
+/// Extract the version string from a sidecar tag name (e.g. "sidecar-v1.0.1" -> "1.0.1").
+fn version_from_sidecar_tag(tag: &str) -> &str {
+    tag.strip_prefix("sidecar-v").unwrap_or(tag)
+}
+
+/// Check if the sidecar binary exists for the currently installed version.
+#[tauri::command]
+pub fn check_sidecar() -> bool {
+    let data_dir = match DATA_DIR.get() {
+        Some(d) => d,
+        None => return false,
+    };
+
+    let version = match read_local_sidecar_version() {
+        Some(v) => v,
+        None => return false,
+    };
+
+    let binary_name = if cfg!(target_os = "windows") {
+        "voice-to-notes-sidecar.exe"
+    } else {
+        "voice-to-notes-sidecar"
+    };
+
+    let extract_dir = data_dir.join(format!("sidecar-{}", version));
+    extract_dir.join(binary_name).exists()
+}
+
+/// Determine the current platform name for asset downloads.
+fn platform_os() -> &'static str {
+    if cfg!(target_os = "windows") {
+        "windows"
+    } else if cfg!(target_os = "macos") {
+        "macos"
+    } else {
+        "linux"
+    }
+}
+
+/// Determine the current architecture name for asset downloads.
+fn platform_arch() -> &'static str {
+    if cfg!(target_arch = "aarch64") {
+        "aarch64"
+    } else {
+        "x86_64"
+    }
+}
+
+/// Download the sidecar binary for the given variant (cpu or cuda).
+#[tauri::command]
+pub async fn download_sidecar(app: AppHandle, variant: String) -> Result<(), String> {
+    let data_dir = DATA_DIR.get().ok_or("App data directory not initialized")?;
+
+    let os = platform_os();
+    let arch = platform_arch();
+    let asset_name = format!("sidecar-{}-{}-{}.zip", os, arch, variant);
+
+    // Fetch the latest sidecar release from Gitea API
+    let client = reqwest::Client::new();
+    let sidecar_release = fetch_latest_sidecar_release(&client).await?;
+
+    let tag = sidecar_release["tag_name"]
+        .as_str()
+        .ok_or("No tag_name in sidecar release")?;
+    let sidecar_version = version_from_sidecar_tag(tag).to_string();
+
+    // Find the matching asset
+    let assets = sidecar_release["assets"]
+        .as_array()
+        .ok_or("No assets found in sidecar release")?;
+
+    let download_url = assets
+        .iter()
+        .find(|a| a["name"].as_str() == Some(&asset_name))
+        .and_then(|a| a["browser_download_url"].as_str())
+        .ok_or_else(|| {
+            format!(
+                "Asset '{}' not found in sidecar release {}",
+                asset_name, tag
+            )
+        })?
+        .to_string();
+
+    // Stream download with progress events
+    let response: reqwest::Response = client
+        .get(&download_url)
+        .send()
+        .await
+        .map_err(|e| format!("Failed to start download: {}", e))?;
+
+    if !response.status().is_success() {
+        return Err(format!("Download failed: HTTP {}", response.status()));
+    }
+
+    let total: u64 = response.content_length().unwrap_or(0);
+    let mut downloaded: u64 = 0;
+    let mut stream = response.bytes_stream();
+
+    let zip_path = data_dir.join("sidecar.zip");
+    let mut file = std::fs::File::create(&zip_path)
+        .map_err(|e| format!("Failed to create zip file: {}", e))?;
+
+    while let Some(chunk) = stream.next().await {
+        let chunk: bytes::Bytes = chunk.map_err(|e| format!("Download stream error: {}", e))?;
+        file.write_all(&chunk)
+            .map_err(|e| format!("Failed to write chunk: {}", e))?;
+        downloaded += chunk.len() as u64;
+        let percent = if total > 0 {
+            (downloaded * 100 / total) as u8
+        } else {
+            0
+        };
+        let _ = app.emit(
+            "sidecar-download-progress",
+            DownloadProgress {
+                downloaded,
+                total,
+                percent,
+            },
+        );
+    }
+
+    // Extract the downloaded zip
+    let extract_dir = data_dir.join(format!("sidecar-{}", sidecar_version));
+    SidecarManager::extract_zip(&zip_path, &extract_dir)?;
+
+    // Make all binaries executable on Unix (sidecar, ffmpeg, ffprobe, etc.)
+    #[cfg(unix)]
+    {
+        use std::os::unix::fs::PermissionsExt;
+        if let Ok(entries) = std::fs::read_dir(&extract_dir) {
+            for entry in entries.flatten() {
+                let path = entry.path();
+                if path.is_file() {
+                    if let Ok(meta) = std::fs::metadata(&path) {
+                        let mut perms = meta.permissions();
+                        perms.set_mode(0o755);
+                        let _ = std::fs::set_permissions(&path, perms);
+                    }
+                }
+            }
+        }
+    }
+
+    // Write the sidecar version file
+    write_local_sidecar_version(&sidecar_version)?;
+
+    // Clean up the zip file and old sidecar versions
+    let _ = std::fs::remove_file(&zip_path);
+    SidecarManager::cleanup_old_sidecars(data_dir, &sidecar_version);
+
+    Ok(())
+}
+
+/// Check if a sidecar update is available.
+#[tauri::command]
+pub async fn check_sidecar_update() -> Result<Option<UpdateInfo>, String> {
+    // If sidecar doesn't exist yet, return None (first launch handled separately)
+    if !check_sidecar() {
+        return Ok(None);
+    }
+
+    let current_version = match read_local_sidecar_version() {
+        Some(v) => v,
+        None => return Ok(None),
+    };
+
+    // Fetch latest sidecar release from Gitea API
+    let client = reqwest::Client::new();
+    let sidecar_release = fetch_latest_sidecar_release(&client).await?;
+
+    let latest_tag = sidecar_release["tag_name"]
+        .as_str()
+        .ok_or("No tag_name in sidecar release")?;
+    let latest_version = version_from_sidecar_tag(latest_tag);
+
+    if latest_version != current_version {
+        Ok(Some(UpdateInfo {
+            current_version,
+            latest_version: latest_version.to_string(),
+        }))
+    } else {
+        Ok(None)
+    }
+}

src-tauri/src/commands/system.rs

@@ -60,3 +60,18 @@ pub fn llama_list_models() -> Value {
 pub fn get_data_dir() -> String {
     LlamaManager::data_dir().to_string_lossy().to_string()
 }
+
+/// Log a message from the frontend to a file for debugging.
+#[tauri::command]
+pub fn log_frontend(level: String, message: String) {
+    use std::io::Write;
+    let log_path = LlamaManager::data_dir().join("frontend.log");
+    if let Ok(mut file) = std::fs::OpenOptions::new()
+        .create(true)
+        .append(true)
+        .open(&log_path)
+    {
+        let timestamp = chrono::Local::now().format("%Y-%m-%d %H:%M:%S");
+        let _ = writeln!(file, "[{timestamp}] [{level}] {message}");
+    }
+}

src-tauri/src/lib.rs

@@ -9,12 +9,16 @@ use tauri::Manager;
 
 use commands::ai::{ai_chat, ai_configure, ai_list_providers};
 use commands::export::export_transcript;
+use commands::media::{check_file_exists, copy_file, create_dir, extract_audio};
 use commands::project::{
     create_project, delete_project, get_project, list_projects, load_project_file,
     load_project_transcript, save_project_file, save_project_transcript, update_segment,
 };
-use commands::settings::{load_settings, save_settings};
+use commands::settings::{load_settings, save_settings, toggle_devtools};
-use commands::system::{get_data_dir, llama_list_models, llama_start, llama_status, llama_stop};
+use commands::sidecar::{check_sidecar, check_sidecar_update, download_sidecar};
+use commands::system::{
+    get_data_dir, llama_list_models, llama_start, llama_status, llama_stop, log_frontend,
+};
 use commands::transcribe::{download_diarize_model, run_pipeline, transcribe_file};
 use state::AppState;
 
@@ -65,6 +69,15 @@ pub fn run() {
             get_data_dir,
             load_settings,
             save_settings,
+            check_sidecar,
+            download_sidecar,
+            check_sidecar_update,
+            log_frontend,
+            toggle_devtools,
+            extract_audio,
+            check_file_exists,
+            copy_file,
+            create_dir,
         ])
         .run(tauri::generate_context!())
         .expect("error while running tauri application");

src-tauri/src/sidecar/mod.rs

@@ -14,7 +14,7 @@ use crate::sidecar::messages::IPCMessage;
 /// Resource directory set by the Tauri app during setup.
 static RESOURCE_DIR: OnceLock<PathBuf> = OnceLock::new();
 /// App data directory for extracting the sidecar archive.
-static DATA_DIR: OnceLock<PathBuf> = OnceLock::new();
+pub(crate) static DATA_DIR: OnceLock<PathBuf> = OnceLock::new();
 
 /// Initialize directories for sidecar resolution.
 /// Must be called from the Tauri setup before any sidecar operations.
@@ -56,10 +56,33 @@ impl SidecarManager {
         cfg!(debug_assertions) || std::env::var("VOICE_TO_NOTES_DEV").is_ok()
     }
 
+    /// Read the locally installed sidecar version from `sidecar-version.txt`.
+    fn read_sidecar_version() -> Result<String, String> {
+        let data_dir = DATA_DIR.get().ok_or("App data directory not initialized")?;
+        let version_file = data_dir.join("sidecar-version.txt");
+        std::fs::read_to_string(&version_file)
+            .map_err(|_| {
+                "Sidecar not installed: sidecar-version.txt not found. Please download the sidecar."
+                    .to_string()
+            })
+            .map(|v| v.trim().to_string())
+            .and_then(|v| {
+                if v.is_empty() {
+                    Err(
+                        "Sidecar version file is empty. Please re-download the sidecar."
+                            .to_string(),
+                    )
+                } else {
+                    Ok(v)
+                }
+            })
+    }
+
     /// Resolve the frozen sidecar binary path (production mode).
     ///
-    /// First checks if the sidecar is already extracted to the app data directory.
+    /// Reads the installed sidecar version from `sidecar-version.txt` and
-    /// If not, looks for `sidecar.zip` in the Tauri resource directory and extracts it.
+    /// looks for the binary in the corresponding `sidecar-{version}` directory.
+    /// If the version file doesn't exist, the sidecar hasn't been downloaded yet.
     fn resolve_sidecar_path() -> Result<PathBuf, String> {
         let binary_name = if cfg!(target_os = "windows") {
             "voice-to-notes-sidecar.exe"
@@ -67,16 +90,15 @@ impl SidecarManager {
             "voice-to-notes-sidecar"
         };
 
-        // Versioned extraction directory prevents stale sidecar after app updates
+        let data_dir = DATA_DIR.get().ok_or("App data directory not initialized")?;
-        let extract_dir = DATA_DIR
+        let current_version = Self::read_sidecar_version()?;
-            .get()
+        let extract_dir = data_dir.join(format!("sidecar-{}", current_version));
-            .ok_or("App data directory not initialized")?
-            .join(format!("sidecar-{}", env!("CARGO_PKG_VERSION")));
 
         let binary_path = extract_dir.join(binary_name);
 
         // Already extracted — use it directly
         if binary_path.exists() {
+            Self::cleanup_old_sidecars(data_dir, &current_version);
             return Ok(binary_path);
         }
 
@@ -91,17 +113,10 @@ impl SidecarManager {
             ));
         }
 
-        // Make executable on Unix
         #[cfg(unix)]
-        {
+        Self::set_executable_permissions(&extract_dir);
-            use std::os::unix::fs::PermissionsExt;
-            if let Ok(meta) = std::fs::metadata(&binary_path) {
-                let mut perms = meta.permissions();
-                perms.set_mode(0o755);
-                let _ = std::fs::set_permissions(&binary_path, perms);
-            }
-        }
 
+        Self::cleanup_old_sidecars(data_dir, &current_version);
         Ok(binary_path)
     }
 
@@ -135,7 +150,7 @@ impl SidecarManager {
     }
 
     /// Extract a zip archive to the given directory.
-    fn extract_zip(zip_path: &Path, dest: &Path) -> Result<(), String> {
+    pub(crate) fn extract_zip(zip_path: &Path, dest: &Path) -> Result<(), String> {
         eprintln!(
             "[sidecar-rs] Extracting sidecar from {} to {}",
             zip_path.display(),
@@ -182,6 +197,62 @@ impl SidecarManager {
         Ok(())
     }
 
+    /// Remove old sidecar-* directories that don't match the current version.
+    /// Called after the current version's sidecar is confirmed ready.
+    /// Set execute permissions on all files in a directory (Unix only).
+    #[cfg(unix)]
+    fn set_executable_permissions(dir: &Path) {
+        use std::os::unix::fs::PermissionsExt;
+        if let Ok(entries) = std::fs::read_dir(dir) {
+            for entry in entries.flatten() {
+                let path = entry.path();
+                if path.is_file() {
+                    if let Ok(meta) = std::fs::metadata(&path) {
+                        let mut perms = meta.permissions();
+                        perms.set_mode(0o755);
+                        let _ = std::fs::set_permissions(&path, perms);
+                    }
+                }
+            }
+        }
+    }
+
+    pub(crate) fn cleanup_old_sidecars(data_dir: &Path, current_version: &str) {
+        let current_dir_name = format!("sidecar-{}", current_version);
+
+        let entries = match std::fs::read_dir(data_dir) {
+            Ok(entries) => entries,
+            Err(e) => {
+                eprintln!("[sidecar-rs] Cannot read data dir for cleanup: {e}");
+                return;
+            }
+        };
+
+        for entry in entries.flatten() {
+            let name = entry.file_name();
+            let name_str = name.to_string_lossy();
+
+            if !name_str.starts_with("sidecar-") {
+                continue;
+            }
+            if *name_str == current_dir_name {
+                continue;
+            }
+            if entry.path().is_dir() {
+                eprintln!(
+                    "[sidecar-rs] Removing old sidecar: {}",
+                    entry.path().display()
+                );
+                if let Err(e) = std::fs::remove_dir_all(entry.path()) {
+                    eprintln!(
+                        "[sidecar-rs] Failed to remove {}: {e}",
+                        entry.path().display()
+                    );
+                }
+            }
+        }
+    }
+
     /// Find a working Python command for the current platform.
     fn find_python_command() -> &'static str {
         if cfg!(target_os = "windows") {
@@ -260,12 +331,40 @@ impl SidecarManager {
         #[cfg(target_os = "windows")]
         cmd.creation_flags(0x08000000);
 
-        let child = cmd
+        match cmd.spawn() {
-            .spawn()
+            Ok(child) => {
-            .map_err(|e| format!("Failed to start sidecar binary: {e}"))?;
+                self.attach(child)?;
-
+                self.wait_for_ready()
-        self.attach(child)?;
+            }
-        self.wait_for_ready()
+            Err(e) if e.raw_os_error() == Some(13) => {
+                // Permission denied — fix permissions and retry once
+                eprintln!("[sidecar-rs] Permission denied, fixing permissions and retrying...");
+                #[cfg(unix)]
+                if let Some(dir) = path.parent() {
+                    Self::set_executable_permissions(dir);
+                }
+                let mut retry_cmd = Command::new(path);
+                retry_cmd
+                    .stdin(Stdio::piped())
+                    .stdout(Stdio::piped())
+                    .stderr(if let Some(data_dir) = DATA_DIR.get() {
+                        let log_path = data_dir.join("sidecar.log");
+                        std::fs::File::create(&log_path)
+                            .map(Stdio::from)
+                            .unwrap_or_else(|_| Stdio::inherit())
+                    } else {
+                        Stdio::inherit()
+                    });
+                #[cfg(target_os = "windows")]
+                retry_cmd.creation_flags(0x08000000);
+                let child = retry_cmd
+                    .spawn()
+                    .map_err(|e| format!("Failed to start sidecar binary after chmod: {e}"))?;
+                self.attach(child)?;
+                self.wait_for_ready()
+            }
+            Err(e) => Err(format!("Failed to start sidecar binary: {e}")),
+        }
     }
 
     /// Spawn the Python sidecar in dev mode (system Python).

src-tauri/tauri.conf.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://schema.tauri.app/config/2",
   "productName": "Voice to Notes",
-  "version": "0.2.5",
+  "version": "0.2.46",
   "identifier": "com.voicetonotes.app",
   "build": {
     "beforeDevCommand": "npm run dev",
@@ -22,7 +22,7 @@
       }
     ],
     "security": {
-      "csp": "default-src 'self'; img-src 'self' asset: https://asset.localhost; media-src 'self' asset: https://asset.localhost; style-src 'self' 'unsafe-inline'",
+      "csp": "default-src 'self' http://tauri.localhost; connect-src ipc: http://ipc.localhost http://asset.localhost https://asset.localhost blob:; img-src 'self' asset: http://asset.localhost https://asset.localhost blob:; media-src 'self' asset: http://asset.localhost https://asset.localhost blob:; style-src 'self' 'unsafe-inline'",
       "assetProtocol": {
         "enable": true,
         "scope": ["**"]
@@ -31,7 +31,7 @@
   },
   "bundle": {
     "active": true,
-    "targets": ["deb", "nsis", "msi", "dmg"],
+    "targets": ["deb", "rpm", "nsis", "msi", "dmg"],
     "icon": [
       "icons/32x32.png",
       "icons/128x128.png",
@@ -42,7 +42,7 @@
     "category": "Utility",
     "shortDescription": "Transcribe audio/video with speaker identification",
     "longDescription": "Voice to Notes is a desktop application that transcribes audio and video recordings with speaker identification, synchronized playback, and AI-powered analysis. Export to SRT, WebVTT, ASS captions, or plain text.",
-    "resources": ["sidecar.zip"],
+    "resources": [],
     "copyright": "Voice to Notes Contributors",
     "license": "MIT",
     "linux": {
@@ -51,6 +51,9 @@
       }
     },
     "windows": {
+      "nsis": {
+        "installerHooks": "nsis-hooks.nsh"
+      },
       "wix": {
         "language": "en-US"
       }

src/lib/components/AIChatPanel.svelte

@@ -1,7 +1,7 @@
 <script lang="ts">
   import { invoke } from '@tauri-apps/api/core';
   import { segments, speakers } from '$lib/stores/transcript';
-  import { settings } from '$lib/stores/settings';
+  import { settings, configureAIProvider } from '$lib/stores/settings';
 
   interface ChatMessage {
     role: 'user' | 'assistant';
@@ -45,22 +45,12 @@
       }));
 
       // Ensure the provider is configured with current credentials before chatting
-      const s = $settings;
+      await configureAIProvider($settings);
-      const configMap: Record<string, Record<string, string>> = {
-        openai: { api_key: s.openai_api_key, model: s.openai_model },
-        anthropic: { api_key: s.anthropic_api_key, model: s.anthropic_model },
-        litellm: { api_key: s.litellm_api_key, api_base: s.litellm_api_base, model: s.litellm_model },
-        local: { model: s.local_model_path, base_url: 'http://localhost:8080' },
-      };
-      const config = configMap[s.ai_provider];
-      if (config) {
-        await invoke('ai_configure', { provider: s.ai_provider, config });
-      }
 
       const result = await invoke<{ response: string }>('ai_chat', {
         messages: chatMessages,
         transcriptContext: getTranscriptContext(),
-        provider: s.ai_provider,
+        provider: $settings.ai_provider,
       });
 
       messages = [...messages, { role: 'assistant', content: result.response }];

src/lib/components/ProgressOverlay.svelte

@@ -4,9 +4,25 @@
     percent?: number;
     stage?: string;
     message?: string;
+    onCancel?: () => void;
   }
 
-  let { visible = false, percent = 0, stage = '', message = '' }: Props = $props();
+  let { visible = false, percent = 0, stage = '', message = '', onCancel }: Props = $props();
+
+  let showConfirm = $state(false);
+
+  function handleCancelClick() {
+    showConfirm = true;
+  }
+
+  function confirmCancel() {
+    showConfirm = false;
+    onCancel?.();
+  }
+
+  function dismissCancel() {
+    showConfirm = false;
+  }
 
   // Pipeline steps in order
   const pipelineSteps = [
@@ -89,6 +105,20 @@
 
       <p class="status-text">{message || 'Please wait...'}</p>
       <p class="hint-text">This may take several minutes for large files</p>
+
+      {#if onCancel && !showConfirm}
+        <button class="cancel-btn" onclick={handleCancelClick}>Cancel</button>
+      {/if}
+
+      {#if showConfirm}
+        <div class="confirm-box">
+          <p class="confirm-text">Processing is incomplete. If you cancel now, the transcription will need to be started over.</p>
+          <div class="confirm-actions">
+            <button class="confirm-keep" onclick={dismissCancel}>Continue Processing</button>
+            <button class="confirm-cancel" onclick={confirmCancel}>Cancel Processing</button>
+          </div>
+        </div>
+      {/if}
     </div>
   </div>
 {/if}
@@ -174,4 +204,62 @@
     font-size: 0.75rem;
     color: #555;
   }
+  .cancel-btn {
+    margin-top: 1.25rem;
+    width: 100%;
+    padding: 0.5rem;
+    background: none;
+    border: 1px solid #4a5568;
+    color: #999;
+    border-radius: 6px;
+    cursor: pointer;
+    font-size: 0.85rem;
+  }
+  .cancel-btn:hover {
+    color: #e0e0e0;
+    border-color: #e94560;
+  }
+  .confirm-box {
+    margin-top: 1.25rem;
+    padding: 0.75rem;
+    background: rgba(233, 69, 96, 0.08);
+    border: 1px solid #e94560;
+    border-radius: 6px;
+  }
+  .confirm-text {
+    margin: 0 0 0.75rem;
+    font-size: 0.8rem;
+    color: #e0e0e0;
+    line-height: 1.4;
+  }
+  .confirm-actions {
+    display: flex;
+    gap: 0.5rem;
+  }
+  .confirm-keep {
+    flex: 1;
+    padding: 0.4rem;
+    background: #0f3460;
+    border: 1px solid #4a5568;
+    color: #e0e0e0;
+    border-radius: 4px;
+    cursor: pointer;
+    font-size: 0.8rem;
+  }
+  .confirm-keep:hover {
+    background: #1a4a7a;
+  }
+  .confirm-cancel {
+    flex: 1;
+    padding: 0.4rem;
+    background: #e94560;
+    border: none;
+    color: white;
+    border-radius: 4px;
+    cursor: pointer;
+    font-size: 0.8rem;
+  }
+  .confirm-cancel:hover {
+    background: #d63851;
+  }
 </style>

src/lib/components/SettingsModal.svelte

@@ -11,7 +11,7 @@
   let { visible, onClose }: Props = $props();
 
   let localSettings = $state<AppSettings>({ ...$settings });
-  let activeTab = $state<'transcription' | 'speakers' | 'ai' | 'local'>('transcription');
+  let activeTab = $state<'transcription' | 'speakers' | 'ai' | 'debug'>('transcription');
   let modelStatus = $state<'idle' | 'downloading' | 'success' | 'error'>('idle');
   let modelError = $state('');
   let revealedFields = $state<Set<string>>(new Set());
@@ -81,8 +81,8 @@
         <button class="tab" class:active={activeTab === 'ai'} onclick={() => activeTab = 'ai'}>
           AI Provider
         </button>
-        <button class="tab" class:active={activeTab === 'local'} onclick={() => activeTab = 'local'}>
+        <button class="tab" class:active={activeTab === 'debug'} onclick={() => activeTab = 'debug'}>
-          Local AI
+          Debug
         </button>
       </div>
 
@@ -181,14 +181,27 @@
           <div class="field">
             <label for="ai-provider">AI Provider</label>
             <select id="ai-provider" bind:value={localSettings.ai_provider}>
-              <option value="local">Local (llama-server)</option>
+              <option value="local">Ollama</option>
               <option value="openai">OpenAI</option>
               <option value="anthropic">Anthropic</option>
               <option value="litellm">OpenAI Compatible</option>
             </select>
           </div>
 
-          {#if localSettings.ai_provider === 'openai'}
+          {#if localSettings.ai_provider === 'local'}
+            <div class="field">
+              <label for="ollama-url">Ollama URL</label>
+              <input id="ollama-url" type="text" bind:value={localSettings.ollama_url} placeholder="http://localhost:11434" />
+            </div>
+            <div class="field">
+              <label for="ollama-model">Model</label>
+              <input id="ollama-model" type="text" bind:value={localSettings.ollama_model} placeholder="llama3.2" />
+            </div>
+            <p class="hint">
+              Install Ollama from ollama.com, then pull a model with <code>ollama pull llama3.2</code>.
+              The app connects via Ollama's OpenAI-compatible API.
+            </p>
+          {:else if localSettings.ai_provider === 'openai'}
             <div class="field">
               <label for="openai-key">OpenAI API Key</label>
               <div class="input-reveal">
@@ -229,19 +242,21 @@
               <input id="litellm-model" type="text" bind:value={localSettings.litellm_model} placeholder="provider/model-name" />
             </div>
           {/if}
-        {:else}
+        {:else if activeTab === 'debug'}
-          <div class="field">
+          <div class="field checkbox">
-            <label for="llama-binary">llama-server Binary Path</label>
+            <label>
-            <input id="llama-binary" type="text" bind:value={localSettings.local_binary_path} placeholder="llama-server" />
+              <input
+                type="checkbox"
+                checked={localSettings.devtools_enabled}
+                onchange={async (e) => {
+                  localSettings.devtools_enabled = (e.target as HTMLInputElement).checked;
+                  await invoke('toggle_devtools', { open: localSettings.devtools_enabled });
+                }}
+              />
+              Enable Developer Tools
+            </label>
+            <p class="hint">Opens the browser inspector for debugging. Changes take effect immediately.</p>
           </div>
-          <div class="field">
-            <label for="llama-model">GGUF Model Path</label>
-            <input id="llama-model" type="text" bind:value={localSettings.local_model_path} placeholder="~/.voicetonotes/models/model.gguf" />
-          </div>
-          <p class="hint">
-            Place GGUF model files in ~/.voicetonotes/models/ for auto-detection.
-            The local AI server uses the OpenAI-compatible API from llama.cpp.
-          </p>
         {/if}
       </div>
 

src/lib/components/SidecarSetup.svelte

@@ -0,0 +1,320 @@
+<script lang="ts">
+  import { invoke } from '@tauri-apps/api/core';
+  import { listen } from '@tauri-apps/api/event';
+  import type { UnlistenFn } from '@tauri-apps/api/event';
+  import { onMount } from 'svelte';
+
+  interface Props {
+    onComplete: () => void;
+  }
+
+  let { onComplete }: Props = $props();
+
+  let variant = $state<'cpu' | 'cuda'>('cpu');
+  let downloading = $state(false);
+  let downloadProgress = $state({ downloaded: 0, total: 0, percent: 0 });
+  let error = $state('');
+  let success = $state(false);
+
+  let unlisten: UnlistenFn | null = null;
+
+  onMount(() => {
+    return () => {
+      unlisten?.();
+    };
+  });
+
+  function formatBytes(bytes: number): string {
+    if (bytes < 1024 * 1024) return `${(bytes / 1024).toFixed(0)} KB`;
+    if (bytes < 1024 * 1024 * 1024) return `${(bytes / (1024 * 1024)).toFixed(0)} MB`;
+    return `${(bytes / (1024 * 1024 * 1024)).toFixed(1)} GB`;
+  }
+
+  async function startDownload() {
+    downloading = true;
+    error = '';
+    success = false;
+
+    unlisten = await listen<{ downloaded: number; total: number; percent: number }>(
+      'sidecar-download-progress',
+      (event) => {
+        downloadProgress = event.payload;
+      }
+    );
+
+    try {
+      await invoke('download_sidecar', { variant });
+      success = true;
+      // Brief pause so the user sees "Complete" before the screen goes away
+      setTimeout(() => {
+        onComplete();
+      }, 800);
+    } catch (err) {
+      error = String(err);
+    } finally {
+      downloading = false;
+      unlisten?.();
+      unlisten = null;
+    }
+  }
+</script>
+
+<div class="setup-overlay">
+  <div class="setup-card">
+    <h1 class="app-title">Voice to Notes</h1>
+    <h2 class="setup-heading">First-Time Setup</h2>
+    <p class="setup-description">
+      Voice to Notes needs to download its AI engine to transcribe audio.
+    </p>
+
+    {#if !downloading && !success}
+      <div class="variant-options">
+        <label class="variant-option" class:selected={variant === 'cpu'}>
+          <input type="radio" name="variant" value="cpu" bind:group={variant} />
+          <div class="variant-info">
+            <span class="variant-label">Standard (CPU)</span>
+            <span class="variant-desc">Works on all computers (~500 MB download)</span>
+          </div>
+        </label>
+        <label class="variant-option" class:selected={variant === 'cuda'}>
+          <input type="radio" name="variant" value="cuda" bind:group={variant} />
+          <div class="variant-info">
+            <span class="variant-label">GPU Accelerated (CUDA)</span>
+            <span class="variant-desc">Faster transcription with NVIDIA GPU (~2 GB download)</span>
+          </div>
+        </label>
+      </div>
+
+      {#if error}
+        <div class="error-box">
+          <p class="error-text">{error}</p>
+          <button class="btn-retry" onclick={startDownload}>Retry</button>
+        </div>
+      {:else}
+        <button class="btn-download" onclick={startDownload}>
+          Download &amp; Install
+        </button>
+      {/if}
+    {:else if downloading}
+      <div class="progress-section">
+        <div class="progress-bar-track">
+          <div class="progress-bar-fill" style="width: {downloadProgress.percent}%"></div>
+        </div>
+        <p class="progress-text">
+          {downloadProgress.percent}% — {formatBytes(downloadProgress.downloaded)} / {formatBytes(downloadProgress.total)}
+        </p>
+        <p class="progress-hint">Downloading {variant === 'cuda' ? 'GPU' : 'CPU'} engine...</p>
+      </div>
+    {:else if success}
+      <div class="success-section">
+        <div class="success-icon">&#10003;</div>
+        <p class="success-text">Setup complete!</p>
+      </div>
+    {/if}
+  </div>
+</div>
+
+<style>
+  .setup-overlay {
+    position: fixed;
+    inset: 0;
+    background: #0a0a23;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    z-index: 10000;
+  }
+
+  .setup-card {
+    background: #16213e;
+    border: 1px solid #2a3a5e;
+    border-radius: 12px;
+    padding: 2.5rem 3rem;
+    max-width: 480px;
+    width: 90vw;
+    color: #e0e0e0;
+    box-shadow: 0 8px 32px rgba(0, 0, 0, 0.5);
+    text-align: center;
+  }
+
+  .app-title {
+    font-size: 1.8rem;
+    margin: 0 0 0.25rem;
+    color: #e94560;
+    font-weight: 700;
+  }
+
+  .setup-heading {
+    font-size: 1.1rem;
+    margin: 0 0 0.75rem;
+    color: #e0e0e0;
+    font-weight: 500;
+  }
+
+  .setup-description {
+    font-size: 0.9rem;
+    color: #b0b0b0;
+    margin: 0 0 1.5rem;
+    line-height: 1.5;
+  }
+
+  .variant-options {
+    display: flex;
+    flex-direction: column;
+    gap: 0.75rem;
+    margin-bottom: 1.5rem;
+    text-align: left;
+  }
+
+  .variant-option {
+    display: flex;
+    align-items: flex-start;
+    gap: 0.75rem;
+    padding: 0.85rem 1rem;
+    border: 1px solid #2a3a5e;
+    border-radius: 8px;
+    cursor: pointer;
+    transition: border-color 0.15s, background 0.15s;
+  }
+
+  .variant-option:hover {
+    border-color: #4a5568;
+    background: rgba(255, 255, 255, 0.02);
+  }
+
+  .variant-option.selected {
+    border-color: #e94560;
+    background: rgba(233, 69, 96, 0.08);
+  }
+
+  .variant-option input[type='radio'] {
+    margin-top: 0.2rem;
+    accent-color: #e94560;
+    flex-shrink: 0;
+  }
+
+  .variant-info {
+    display: flex;
+    flex-direction: column;
+    gap: 0.2rem;
+  }
+
+  .variant-label {
+    font-size: 0.9rem;
+    font-weight: 500;
+    color: #e0e0e0;
+  }
+
+  .variant-desc {
+    font-size: 0.78rem;
+    color: #888;
+  }
+
+  .btn-download {
+    background: #e94560;
+    border: none;
+    color: white;
+    padding: 0.7rem 1.5rem;
+    border-radius: 6px;
+    cursor: pointer;
+    font-size: 0.9rem;
+    font-weight: 500;
+    width: 100%;
+    transition: background 0.15s;
+  }
+
+  .btn-download:hover {
+    background: #d63851;
+  }
+
+  .progress-section {
+    margin-top: 0.5rem;
+  }
+
+  .progress-bar-track {
+    width: 100%;
+    height: 8px;
+    background: #1a1a2e;
+    border-radius: 4px;
+    overflow: hidden;
+    border: 1px solid #2a3a5e;
+  }
+
+  .progress-bar-fill {
+    height: 100%;
+    background: #e94560;
+    border-radius: 4px;
+    transition: width 0.3s ease;
+  }
+
+  .progress-text {
+    margin: 0.75rem 0 0;
+    font-size: 0.85rem;
+    color: #e0e0e0;
+    font-variant-numeric: tabular-nums;
+  }
+
+  .progress-hint {
+    margin: 0.35rem 0 0;
+    font-size: 0.78rem;
+    color: #888;
+  }
+
+  .error-box {
+    background: rgba(233, 69, 96, 0.1);
+    border: 1px solid rgba(233, 69, 96, 0.3);
+    border-radius: 8px;
+    padding: 1rem;
+  }
+
+  .error-text {
+    color: #e94560;
+    font-size: 0.85rem;
+    margin: 0 0 0.75rem;
+    word-break: break-word;
+    line-height: 1.4;
+  }
+
+  .btn-retry {
+    background: #e94560;
+    border: none;
+    color: white;
+    padding: 0.5rem 1.25rem;
+    border-radius: 6px;
+    cursor: pointer;
+    font-size: 0.85rem;
+    font-weight: 500;
+  }
+
+  .btn-retry:hover {
+    background: #d63851;
+  }
+
+  .success-section {
+    display: flex;
+    flex-direction: column;
+    align-items: center;
+    gap: 0.5rem;
+    padding: 1rem 0;
+  }
+
+  .success-icon {
+    width: 48px;
+    height: 48px;
+    border-radius: 50%;
+    background: rgba(78, 205, 196, 0.15);
+    color: #4ecdc4;
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    font-size: 1.5rem;
+    font-weight: 700;
+  }
+
+  .success-text {
+    color: #4ecdc4;
+    font-size: 1rem;
+    margin: 0;
+    font-weight: 500;
+  }
+</style>

src/lib/components/TranscriptEditor.svelte

@@ -272,7 +272,9 @@
 <style>
   .transcript-editor {
     flex: 1;
+    min-width: 0;
     overflow-y: auto;
+    overflow-x: hidden;
     padding: 1rem;
     background: #16213e;
     border-radius: 8px;
@@ -319,6 +321,7 @@
   .segment-text {
     line-height: 1.6;
     padding-left: 0.75rem;
+    white-space: pre-wrap;
     word-wrap: break-word;
     overflow-wrap: break-word;
   }

src/lib/components/WaveformPlayer.svelte

@@ -57,6 +57,12 @@
       isReady = false;
     });
 
+    wavesurfer.on('error', (err: Error) => {
+      console.error('[voice-to-notes] WaveSurfer error:', err);
+      isLoading = false;
+      loadError = 'Failed to load audio';
+    });
+
     if (audioUrl) {
       loadAudio(audioUrl);
     }

src/lib/stores/settings.ts

@@ -10,14 +10,15 @@ export interface AppSettings {
   litellm_model: string;
   litellm_api_key: string;
   litellm_api_base: string;
-  local_model_path: string;
+  ollama_url: string;
-  local_binary_path: string;
+  ollama_model: string;
   transcription_model: string;
   transcription_device: string;
   transcription_language: string;
   skip_diarization: boolean;
   hf_token: string;
   num_speakers: number | null;
+  devtools_enabled: boolean;
 }
 
 const defaults: AppSettings = {
@@ -29,14 +30,15 @@ const defaults: AppSettings = {
   litellm_model: 'gpt-4o-mini',
   litellm_api_key: '',
   litellm_api_base: '',
-  local_model_path: '',
+  ollama_url: 'http://localhost:11434',
-  local_binary_path: 'llama-server',
+  ollama_model: 'llama3.2',
   transcription_model: 'base',
   transcription_device: 'cpu',
   transcription_language: '',
   skip_diarization: false,
   hf_token: '',
   num_speakers: null,
+  devtools_enabled: false,
 };
 
 export const settings = writable<AppSettings>({ ...defaults });
@@ -50,23 +52,27 @@ export async function loadSettings(): Promise<void> {
   }
 }
 
-export async function saveSettings(s: AppSettings): Promise<void> {
+export async function configureAIProvider(s: AppSettings): Promise<void> {
-  settings.set(s);
-  await invoke('save_settings', { settings: s });
-
-  // Configure the AI provider in the Python sidecar
   const configMap: Record<string, Record<string, string>> = {
     openai: { api_key: s.openai_api_key, model: s.openai_model },
     anthropic: { api_key: s.anthropic_api_key, model: s.anthropic_model },
     litellm: { api_key: s.litellm_api_key, api_base: s.litellm_api_base, model: s.litellm_model },
-    local: { model: s.local_model_path, base_url: 'http://localhost:8080' },
+    local: { model: s.ollama_model, base_url: s.ollama_url.replace(/\/+$/, '') + '/v1' },
   };
   const config = configMap[s.ai_provider];
   if (config) {
     try {
       await invoke('ai_configure', { provider: s.ai_provider, config });
     } catch {
-      // Sidecar may not be running yet — provider will be configured on first use
+      // Sidecar may not be running yet
     }
   }
 }
+
+export async function saveSettings(s: AppSettings): Promise<void> {
+  settings.set(s);
+  await invoke('save_settings', { settings: s });
+
+  // Configure the AI provider in the Python sidecar
+  await configureAIProvider(s);
+}

src/routes/+page.svelte

@@ -8,8 +8,9 @@
   import AIChatPanel from '$lib/components/AIChatPanel.svelte';
   import ProgressOverlay from '$lib/components/ProgressOverlay.svelte';
   import SettingsModal from '$lib/components/SettingsModal.svelte';
+  import SidecarSetup from '$lib/components/SidecarSetup.svelte';
   import { segments, speakers } from '$lib/stores/transcript';
-  import { settings, loadSettings } from '$lib/stores/settings';
+  import { settings, loadSettings, configureAIProvider } from '$lib/stores/settings';
   import type { Segment, Speaker } from '$lib/types/transcript';
   import { onMount, tick } from 'svelte';
 
@@ -18,13 +19,65 @@
   let audioUrl = $state('');
   let showSettings = $state(false);
 
+  // Sidecar state
+  let sidecarReady = $state(false);
+  let sidecarChecked = $state(false);
+
+  // Sidecar update state
+  let sidecarUpdate = $state<{ current_version: string; latest_version: string } | null>(null);
+  let showUpdateDownload = $state(false);
+  let updateDismissed = $state(false);
+
   // Project management state
   let currentProjectPath = $state<string | null>(null);
   let currentProjectName = $state('');
+  let projectIsV2 = $state(false);
   let audioFilePath = $state('');
+  let audioWavPath = $state('');
+
+  async function checkSidecar() {
+    try {
+      const ready = await invoke<boolean>('check_sidecar');
+      sidecarReady = ready;
+    } catch {
+      sidecarReady = false;
+    }
+    sidecarChecked = true;
+  }
+
+  async function checkSidecarUpdate() {
+    try {
+      const update = await invoke<{ current_version: string; latest_version: string } | null>('check_sidecar_update');
+      sidecarUpdate = update;
+    } catch {
+      // Silently ignore update check failures
+    }
+  }
+
+  function handleSidecarSetupComplete() {
+    sidecarReady = true;
+    configureAIProvider($settings);
+    checkSidecarUpdate();
+  }
+
+  function handleUpdateComplete() {
+    showUpdateDownload = false;
+    sidecarUpdate = null;
+  }
 
   onMount(() => {
-    loadSettings();
+    loadSettings().then(() => {
+      // Restore devtools state from settings
+      if ($settings.devtools_enabled) {
+        invoke('toggle_devtools', { open: true });
+      }
+    });
+    checkSidecar().then(() => {
+      if (sidecarReady) {
+        configureAIProvider($settings);
+        checkSidecarUpdate();
+      }
+    });
 
     // Global keyboard shortcuts
     function handleKeyDown(e: KeyboardEvent) {
@@ -68,25 +121,32 @@
     };
   });
   let isTranscribing = $state(false);
+  let transcriptionCancelled = $state(false);
   let transcriptionProgress = $state(0);
   let transcriptionStage = $state('');
   let transcriptionMessage = $state('');
+  let extractingAudio = $state(false);
+
+  function handleCancelProcessing() {
+    transcriptionCancelled = true;
+    isTranscribing = false;
+    transcriptionProgress = 0;
+    transcriptionStage = '';
+    transcriptionMessage = '';
+    // Clear any partial results
+    segments.set([]);
+    speakers.set([]);
+  }
 
   // Speaker color palette for auto-assignment
   const speakerColors = ['#e94560', '#4ecdc4', '#ffe66d', '#a8e6cf', '#ff8b94', '#c7ceea', '#ffd93d', '#6bcb77'];
 
-  async function saveProject() {
+  function buildProjectData(projectName: string) {
-    const defaultName = currentProjectName || 'Untitled';
+    return {
-    const outputPath = await save({
+      version: 2,
-      defaultPath: `${defaultName}.vtn`,
+      name: projectName,
-      filters: [{ name: 'Voice to Notes Project', extensions: ['vtn'] }],
+      source_file: audioFilePath,
-    });
+      audio_wav: 'audio.wav',
-    if (!outputPath) return;
-
-    const projectData = {
-      version: 1,
-      name: outputPath.split(/[\\/]/).pop()?.replace('.vtn', '') || defaultName,
-      audio_file: audioFilePath,
       created_at: new Date().toISOString(),
       segments: $segments.map(seg => {
         const speaker = $speakers.find(s => s.id === seg.speaker_id);
@@ -110,17 +170,75 @@
         color: s.color || '#e94560',
       })),
     };
+  }
+
+  /** Save to a specific folder — creates .vtn + audio.wav inside it. */
+  async function saveToFolder(folderPath: string): Promise<boolean> {
+    const projectName = folderPath.split(/[\\/]/).pop() || currentProjectName || 'Untitled';
+    const vtnPath = `${folderPath}/${projectName}.vtn`;
+    const wavPath = `${folderPath}/audio.wav`;
+    const projectData = buildProjectData(projectName);
 
     try {
-      await invoke('save_project_file', { path: outputPath, project: projectData });
+      await invoke('create_dir', { path: folderPath });
-      currentProjectPath = outputPath;
+      if (audioWavPath && audioWavPath !== wavPath) {
-      currentProjectName = projectData.name;
+        await invoke('copy_file', { src: audioWavPath, dst: wavPath });
+        audioWavPath = wavPath;
+      }
+      await invoke('save_project_file', { path: vtnPath, project: projectData });
+      currentProjectPath = vtnPath;
+      currentProjectName = projectName;
+      projectIsV2 = true;
+      return true;
     } catch (err) {
       console.error('Failed to save project:', err);
       alert(`Failed to save: ${err}`);
+      return false;
     }
   }
 
+  async function saveProject() {
+    // Already saved as v2 folder — save in place
+    if (currentProjectPath && projectIsV2) {
+      const folderPath = currentProjectPath.replace(/[\\/][^\\/]+$/, '');
+      await saveToFolder(folderPath);
+      return;
+    }
+
+    // V1 project opened — migrate to folder structure
+    if (currentProjectPath && !projectIsV2) {
+      const oldVtnDir = currentProjectPath.replace(/[\\/][^\\/]+$/, '');
+      const projectName = currentProjectPath.split(/[\\/]/).pop()?.replace(/\.vtn$/i, '') || 'Untitled';
+      const folderPath = `${oldVtnDir}/${projectName}`;
+      const success = await saveToFolder(folderPath);
+      if (success) {
+        // Optionally remove the old .vtn file
+        try {
+          // Leave old file — user can delete manually
+        } catch {}
+      }
+      return;
+    }
+
+    // Never saved — pick a folder
+    await saveProjectAs();
+  }
+
+  async function saveProjectAs() {
+    // Use save dialog so the user can type a new project name.
+    // The chosen path is treated as the project folder (created if needed).
+    const defaultName = currentProjectName || 'Untitled';
+    const chosenPath = await save({
+      defaultPath: defaultName,
+      title: 'Save Project — enter a project name',
+    });
+    if (!chosenPath) return;
+
+    // Strip any file extension the user may have typed (e.g. ".vtn")
+    const folderPath = chosenPath.replace(/\.[^.\\/]+$/, '');
+    await saveToFolder(folderPath);
+  }
+
   async function openProject() {
     const filePath = await open({
       filters: [{ name: 'Voice to Notes Project', extensions: ['vtn'] }],
@@ -130,9 +248,11 @@
 
     try {
       const project = await invoke<{
-        version: number;
+        version?: number;
         name: string;
-        audio_file: string;
+        audio_file?: string;
+        source_file?: string;
+        audio_wav?: string;
         segments: Array<{
           text: string;
           start_ms: number;
@@ -182,10 +302,135 @@
       }));
       segments.set(newSegments);
 
-      // Load audio
+      // Determine the directory the .vtn file is in
-      audioFilePath = project.audio_file;
+      const vtnDir = (filePath as string).replace(/[\\/][^\\/]+$/, '');
-      audioUrl = convertFileSrc(project.audio_file);
+      const version = project.version ?? 1;
-      waveformPlayer?.loadAudio(audioUrl);
+      projectIsV2 = version >= 2;
+
+      // Resolve audio for wavesurfer playback
+      if (version >= 2) {
+        // Version 2: audio_wav is relative to the .vtn directory, source_file is the original import path
+        audioFilePath = project.source_file || '';
+        const wavRelative = project.audio_wav || 'audio.wav';
+        const resolvedWav = `${vtnDir}/${wavRelative}`;
+
+        const wavExists = await invoke<boolean>('check_file_exists', { path: resolvedWav });
+        if (wavExists) {
+          audioWavPath = resolvedWav;
+          audioUrl = convertFileSrc(resolvedWav);
+          waveformPlayer?.loadAudio(audioUrl);
+        } else {
+          // WAV missing — try re-extracting from the original source file
+          const sourceExists = audioFilePath ? await invoke<boolean>('check_file_exists', { path: audioFilePath }) : false;
+          if (sourceExists) {
+            extractingAudio = true;
+            await tick();
+            try {
+              const outputPath = `${vtnDir}/${wavRelative}`;
+              const wavPath = await invoke<string>('extract_audio', { filePath: audioFilePath, outputPath });
+              audioWavPath = wavPath;
+              audioUrl = convertFileSrc(wavPath);
+              waveformPlayer?.loadAudio(audioUrl);
+            } catch (err) {
+              console.error('Failed to re-extract audio:', err);
+              alert(`Failed to re-extract audio: ${err}`);
+            } finally {
+              extractingAudio = false;
+            }
+          } else {
+            // Both missing — ask user to locate the file
+            const shouldRelink = confirm(
+              'The audio file for this project could not be found.\n\n' +
+              `Original source: ${audioFilePath || '(unknown)'}\n\n` +
+              'Would you like to locate the file?'
+            );
+            if (shouldRelink) {
+              const newPath = await open({
+                multiple: false,
+                filters: [{
+                  name: 'Audio/Video',
+                  extensions: ['mp3', 'wav', 'flac', 'ogg', 'm4a', 'aac', 'wma',
+                               'mp4', 'mkv', 'avi', 'mov', 'webm'],
+                }],
+              });
+              if (newPath) {
+                audioFilePath = newPath;
+                extractingAudio = true;
+                await tick();
+                try {
+                  const outputPath = `${vtnDir}/${wavRelative}`;
+                  const wavPath = await invoke<string>('extract_audio', { filePath: newPath, outputPath });
+                  audioWavPath = wavPath;
+                  audioUrl = convertFileSrc(wavPath);
+                  waveformPlayer?.loadAudio(audioUrl);
+                } catch (err) {
+                  console.error('Failed to extract audio from re-linked file:', err);
+                  alert(`Failed to extract audio: ${err}`);
+                } finally {
+                  extractingAudio = false;
+                }
+              }
+            }
+          }
+        }
+      } else {
+        // Version 1 (legacy): audio_file is the source path
+        const sourceFile = project.audio_file || '';
+        audioFilePath = sourceFile;
+
+        const sourceExists = sourceFile ? await invoke<boolean>('check_file_exists', { path: sourceFile }) : false;
+        if (sourceExists) {
+          // Extract WAV next to the .vtn file for playback
+          extractingAudio = true;
+          await tick();
+          try {
+            const outputPath = `${vtnDir}/audio.wav`;
+            const wavPath = await invoke<string>('extract_audio', { filePath: sourceFile, outputPath });
+            audioWavPath = wavPath;
+            audioUrl = convertFileSrc(wavPath);
+            waveformPlayer?.loadAudio(audioUrl);
+          } catch (err) {
+            console.error('Failed to extract audio:', err);
+            alert(`Failed to extract audio: ${err}`);
+          } finally {
+            extractingAudio = false;
+          }
+        } else {
+          // Source missing — ask user to locate the file
+          const shouldRelink = confirm(
+            'The audio file for this project could not be found.\n\n' +
+            `Original path: ${sourceFile || '(unknown)'}\n\n` +
+            'Would you like to locate the file?'
+          );
+          if (shouldRelink) {
+            const newPath = await open({
+              multiple: false,
+              filters: [{
+                name: 'Audio/Video',
+                extensions: ['mp3', 'wav', 'flac', 'ogg', 'm4a', 'aac', 'wma',
+                             'mp4', 'mkv', 'avi', 'mov', 'webm'],
+              }],
+            });
+            if (newPath) {
+              audioFilePath = newPath;
+              extractingAudio = true;
+              await tick();
+              try {
+                const outputPath = `${vtnDir}/audio.wav`;
+                const wavPath = await invoke<string>('extract_audio', { filePath: newPath, outputPath });
+                audioWavPath = wavPath;
+                audioUrl = convertFileSrc(wavPath);
+                waveformPlayer?.loadAudio(audioUrl);
+              } catch (err) {
+                console.error('Failed to extract audio from re-linked file:', err);
+                alert(`Failed to extract audio: ${err}`);
+              } finally {
+                extractingAudio = false;
+              }
+            }
+          }
+        }
+      }
 
       currentProjectPath = filePath as string;
       currentProjectName = project.name;
@@ -216,9 +461,35 @@
     });
     if (!filePath) return;
 
-    // Track the original file path and convert to asset URL for wavesurfer
+    // Always extract audio to WAV for wavesurfer playback
+    extractingAudio = true;
+    await tick();
+    try {
+      const wavPath = await invoke<string>('extract_audio', { filePath });
+      audioWavPath = wavPath;
+    } catch (err) {
+      console.error('[voice-to-notes] Failed to extract audio:', err);
+      const msg = String(err);
+      if (msg.includes('ffmpeg not found')) {
+        alert(
+          'FFmpeg is required to extract audio.\n\n' +
+          'Install FFmpeg:\n' +
+          '  Windows: winget install ffmpeg\n' +
+          '  macOS: brew install ffmpeg\n' +
+          '  Linux: sudo apt install ffmpeg\n\n' +
+          'Then restart Voice to Notes and try again.'
+        );
+      } else {
+        alert(`Failed to extract audio: ${msg}`);
+      }
+      return;
+    } finally {
+      extractingAudio = false;
+    }
+
+    // Track the original file path for the sidecar (it does its own conversion)
     audioFilePath = filePath;
-    audioUrl = convertFileSrc(filePath);
+    audioUrl = convertFileSrc(audioWavPath);
     waveformPlayer?.loadAudio(audioUrl);
 
     // Clear previous results
@@ -227,6 +498,7 @@
 
     // Start pipeline (transcription + diarization)
     isTranscribing = true;
+    transcriptionCancelled = false;
     transcriptionProgress = 0;
     transcriptionStage = 'Starting...';
     transcriptionMessage = 'Initializing pipeline...';
@@ -337,6 +609,9 @@
         numSpeakers: $settings.num_speakers && $settings.num_speakers > 0 ? $settings.num_speakers : undefined,
       });
 
+      // If cancelled while processing, discard results
+      if (transcriptionCancelled) return;
+
       // Create speaker entries from pipeline result
       const newSpeakers: Speaker[] = (result.speakers || []).map((label, idx) => ({
         id: `speaker-${idx}`,
@@ -443,14 +718,31 @@
   }
 </script>
 
-{#if !appReady}
+{#if !appReady || !sidecarChecked}
   <div class="splash-screen">
     <h1 class="splash-title">Voice to Notes</h1>
     <p class="splash-subtitle">Loading...</p>
     <div class="splash-spinner"></div>
   </div>
+{:else if sidecarChecked && !sidecarReady && !showUpdateDownload}
+  <SidecarSetup onComplete={handleSidecarSetupComplete} />
+{:else if showUpdateDownload}
+  <SidecarSetup onComplete={handleUpdateComplete} />
 {:else}
   <div class="app-shell">
+  {#if sidecarUpdate && !updateDismissed}
+    <div class="update-banner">
+      <span class="update-text">
+        Sidecar update available (v{sidecarUpdate.current_version} &rarr; v{sidecarUpdate.latest_version})
+      </span>
+      <button class="update-btn" onclick={() => showUpdateDownload = true}>
+        Update
+      </button>
+      <button class="update-dismiss" onclick={() => updateDismissed = true} title="Dismiss">
+        &times;
+      </button>
+    </div>
+  {/if}
   <div class="app-header">
     <div class="header-actions">
       <button class="settings-btn" onclick={openProject} disabled={isTranscribing}>
@@ -458,7 +750,10 @@
       </button>
       {#if $segments.length > 0}
         <button class="settings-btn" onclick={saveProject}>
-          Save Project
+          Save
+        </button>
+        <button class="settings-btn" onclick={saveProjectAs}>
+          Save As
         </button>
       {/if}
       <button class="import-btn" onclick={handleFileImport} disabled={isTranscribing}>
@@ -507,8 +802,18 @@
     percent={transcriptionProgress}
     stage={transcriptionStage}
     message={transcriptionMessage}
+    onCancel={handleCancelProcessing}
   />
 
+  {#if extractingAudio}
+    <div class="extraction-overlay">
+      <div class="extraction-card">
+        <div class="extraction-spinner"></div>
+        <p>Extracting audio...</p>
+      </div>
+    </div>
+  {/if}
+
   <SettingsModal
     visible={showSettings}
     onClose={() => showSettings = false}
@@ -674,4 +979,80 @@
   @keyframes spin {
     to { transform: rotate(360deg); }
   }
+
+  /* Sidecar update banner */
+  .update-banner {
+    display: flex;
+    align-items: center;
+    gap: 0.75rem;
+    padding: 0.5rem 1rem;
+    background: rgba(78, 205, 196, 0.1);
+    border-bottom: 1px solid rgba(78, 205, 196, 0.25);
+    color: #e0e0e0;
+    font-size: 0.85rem;
+  }
+  .update-text {
+    flex: 1;
+    color: #b0b0b0;
+  }
+  .update-btn {
+    background: #4ecdc4;
+    border: none;
+    color: #0a0a23;
+    padding: 0.3rem 0.85rem;
+    border-radius: 4px;
+    cursor: pointer;
+    font-size: 0.8rem;
+    font-weight: 600;
+  }
+  .update-btn:hover {
+    background: #3dbdb5;
+  }
+  .update-dismiss {
+    background: none;
+    border: none;
+    color: #888;
+    font-size: 1.1rem;
+    cursor: pointer;
+    padding: 0.1rem 0.3rem;
+    line-height: 1;
+  }
+  .update-dismiss:hover {
+    color: #e0e0e0;
+  }
+
+  /* Audio extraction overlay */
+  .extraction-overlay {
+    position: fixed;
+    inset: 0;
+    background: rgba(0, 0, 0, 0.8);
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    z-index: 9999;
+  }
+  .extraction-card {
+    background: #16213e;
+    padding: 2rem 2.5rem;
+    border-radius: 12px;
+    color: #e0e0e0;
+    border: 1px solid #2a3a5e;
+    box-shadow: 0 8px 32px rgba(0, 0, 0, 0.5);
+    display: flex;
+    flex-direction: column;
+    align-items: center;
+    gap: 1rem;
+  }
+  .extraction-card p {
+    margin: 0;
+    font-size: 1rem;
+  }
+  .extraction-spinner {
+    width: 32px;
+    height: 32px;
+    border: 3px solid #2a3a5e;
+    border-top-color: #e94560;
+    border-radius: 50%;
+    animation: spin 0.8s linear infinite;
+  }
 </style>