chore: bump sidecar version to 1.0.7 [skip ci]

CUDA sidecars are ~2GB and too slow to upload from the Windows runner.
Cloud (Deepgram) provides faster transcription anyway. Removed:

- CUDA build steps from Windows and Linux sidecar workflows
- CUDA option from the SidecarSetup download screen

Remaining sidecar variants:
- Cloud (Deepgram): ~50 MB - recommended for most users
- Local CPU: ~500 MB - for offline/privacy use

CUDA can be revisited once the managed Deepgram service is ready.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

.gitea/workflows/build-app-linux.yml

@@ -13,23 +13,15 @@ jobs:
     runs-on: ubuntu-latest
     env:
       NODE_VERSION: "20"
+      RELEASE_TAG: "${{ inputs.tag }}"
     steps:
-      - name: Determine tag
+      - name: Show tag
-        id: tag
         run: |
-          TAG="${{ inputs.tag }}"
+          echo "Building for tag: ${RELEASE_TAG}"
-          if [ -z "$TAG" ]; then
-            TAG="${{ github.event.inputs.tag }}"
-          fi
-          if [ -z "$TAG" ]; then
-            TAG=$(git ls-remote --tags --sort=-v:refname origin 'refs/tags/v*' | head -1 | sed 's|.*refs/tags/||')
-          fi
-          echo "Building for tag: ${TAG}"
-          echo "tag=${TAG}" >> $GITHUB_OUTPUT
 
       - uses: actions/checkout@v4
         with:
-          ref: ${{ steps.tag.outputs.tag }}
+          ref: ${{ inputs.tag }}
 
       - name: Set up Node.js
         uses: actions/setup-node@v4
@@ -58,7 +50,7 @@ jobs:
         run: |
           sudo apt-get install -y jq
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
-          TAG="${{ steps.tag.outputs.tag }}"
+          TAG="${RELEASE_TAG}"
           echo "Release tag: ${TAG}"
 
           echo "Waiting for release ${TAG} to be available..."

.gitea/workflows/build-app-macos.yml

@@ -13,23 +13,15 @@ jobs:
     runs-on: macos-latest
     env:
       NODE_VERSION: "20"
+      RELEASE_TAG: "${{ inputs.tag }}"
     steps:
-      - name: Determine tag
+      - name: Show tag
-        id: tag
         run: |
-          TAG="${{ inputs.tag }}"
+          echo "Building for tag: ${RELEASE_TAG}"
-          if [ -z "$TAG" ]; then
-            TAG="${{ github.event.inputs.tag }}"
-          fi
-          if [ -z "$TAG" ]; then
-            TAG=$(git ls-remote --tags --sort=-v:refname origin 'refs/tags/v*' | head -1 | sed 's|.*refs/tags/||')
-          fi
-          echo "Building for tag: ${TAG}"
-          echo "tag=${TAG}" >> $GITHUB_OUTPUT
 
       - uses: actions/checkout@v4
         with:
-          ref: ${{ steps.tag.outputs.tag }}
+          ref: ${{ inputs.tag }}
 
       - name: Set up Node.js
         uses: actions/setup-node@v4
@@ -56,7 +48,7 @@ jobs:
         run: |
           which jq || brew install jq
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
-          TAG="${{ steps.tag.outputs.tag }}"
+          TAG="${RELEASE_TAG}"
           echo "Release tag: ${TAG}"
 
           echo "Waiting for release ${TAG} to be available..."

.gitea/workflows/build-app-windows.yml

@@ -15,7 +15,7 @@ jobs:
     name: Build App (Windows)
     runs-on: windows-latest
     env:
-      RELEASE_TAG: ${{ inputs.tag }}
+      RELEASE_TAG: "${{ inputs.tag }}"
     steps:
       - name: Show tag
         shell: powershell

.gitea/workflows/build-sidecar-cloud.yml

@@ -0,0 +1,229 @@
+name: Build Sidecar (Cloud)
+
+on:
+  workflow_dispatch:
+    inputs:
+      tag:
+        description: 'Sidecar release tag to build (e.g. sidecar-v1.0.5)'
+        required: true
+
+jobs:
+  build-cloud-linux:
+    name: Build Cloud Sidecar (Linux)
+    runs-on: ubuntu-latest
+    env:
+      PYTHON_VERSION: "3.11"
+      RELEASE_TAG: "${{ inputs.tag }}"
+    steps:
+      - name: Show tag
+        run: |
+          echo "Building cloud sidecar for tag ${RELEASE_TAG}"
+
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ inputs.tag }}
+
+      - name: Install uv
+        run: |
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          echo "$HOME/.local/bin" >> $GITHUB_PATH
+
+      - name: Set up Python
+        run: uv python install ${{ env.PYTHON_VERSION }}
+
+      - name: Install system dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y portaudio19-dev
+
+      - name: Build cloud sidecar
+        env:
+          UV_NO_SOURCES: "1"
+        run: |
+          uv venv
+          uv pip install pyinstaller numpy sounddevice fastapi uvicorn websockets pydantic requests pyyaml packaging
+          .venv/bin/pyinstaller local-transcription-cloud.spec
+
+      - name: Package
+        run: |
+          cd dist/local-transcription-backend && zip -r ../../sidecar-linux-x86_64-cloud.zip .
+
+      - name: Upload to release
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          sudo apt-get install -y jq
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+          TAG="${RELEASE_TAG}"
+
+          for i in $(seq 1 30); do
+            RELEASE_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/tags/${TAG}" | jq -r '.id // empty')
+            if [ -n "${RELEASE_ID}" ] && [ "${RELEASE_ID}" != "null" ]; then
+              echo "Found release ${TAG} (ID: ${RELEASE_ID})"
+              break
+            fi
+            echo "Attempt ${i}/30: waiting for release..."
+            sleep 10
+          done
+
+          if [ -z "${RELEASE_ID}" ] || [ "${RELEASE_ID}" = "null" ]; then
+            echo "ERROR: Release not found"; exit 1
+          fi
+
+          for file in sidecar-*-cloud.zip; do
+            filename=$(basename "$file")
+            ASSET_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/${RELEASE_ID}/assets" | jq -r ".[] | select(.name == \"${filename}\") | .id // empty")
+            [ -n "${ASSET_ID}" ] && curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" "${REPO_API}/releases/${RELEASE_ID}/assets/${ASSET_ID}"
+            curl -s -o /dev/null -w "Upload ${filename}: HTTP %{http_code}\n" -X POST \
+              -H "Authorization: token ${BUILD_TOKEN}" -H "Content-Type: application/octet-stream" \
+              -T "$file" "${REPO_API}/releases/${RELEASE_ID}/assets?name=${filename}"
+          done
+
+  build-cloud-windows:
+    name: Build Cloud Sidecar (Windows)
+    runs-on: windows-latest
+    env:
+      PYTHON_VERSION: "3.11"
+      RELEASE_TAG: "${{ inputs.tag }}"
+    steps:
+      - name: Show tag
+        shell: powershell
+        run: Write-Host "Building cloud sidecar for tag $env:RELEASE_TAG"
+
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ inputs.tag }}
+
+      - name: Install uv
+        shell: powershell
+        run: |
+          if (Get-Command uv -ErrorAction SilentlyContinue) {
+            Write-Host "uv already installed"
+          } else {
+            irm https://astral.sh/uv/install.ps1 | iex
+            $uvPaths = @("$env:USERPROFILE\.local\bin", "$env:USERPROFILE\.cargo\bin", "$env:LOCALAPPDATA\uv\bin")
+            foreach ($p in $uvPaths) { if (Test-Path $p) { echo $p | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append } }
+          }
+
+      - name: Set up Python
+        shell: powershell
+        run: uv python install ${{ env.PYTHON_VERSION }}
+
+      - name: Build cloud sidecar
+        shell: powershell
+        env:
+          UV_NO_SOURCES: "1"
+        run: |
+          uv venv
+          uv pip install pyinstaller numpy sounddevice fastapi uvicorn websockets pydantic requests pyyaml packaging
+          .venv\Scripts\pyinstaller.exe local-transcription-cloud.spec
+
+      - name: Package
+        shell: powershell
+        run: |
+          if (-not (Get-Command 7z -ErrorAction SilentlyContinue)) { choco install 7zip -y }
+          7z a -tzip -mx=5 sidecar-windows-x86_64-cloud.zip .\dist\local-transcription-backend\*
+
+      - name: Upload to 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 = $env:RELEASE_TAG
+
+          $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 release $TAG (ID: $RELEASE_ID)"; break }
+            } catch {}
+            Write-Host "Attempt ${i}/30: waiting..."; Start-Sleep -Seconds 10
+          }
+          if (-not $RELEASE_ID) { Write-Host "ERROR: Release not found"; exit 1 }
+
+          Get-ChildItem -Path . -Filter "sidecar-*-cloud.zip" | ForEach-Object {
+            $fn = $_.Name; $enc = [System.Uri]::EscapeDataString($fn)
+            try {
+              $assets = Invoke-RestMethod -Uri "$REPO_API/releases/$RELEASE_ID/assets" -Headers $Headers
+              $existing = $assets | Where-Object { $_.name -eq $fn }
+              if ($existing) { Invoke-RestMethod -Uri "$REPO_API/releases/$RELEASE_ID/assets/$($existing.id)" -Method Delete -Headers $Headers }
+            } catch {}
+            curl.exe --fail -s -X POST -H "Authorization: token $env:BUILD_TOKEN" -H "Content-Type: application/octet-stream" -T "$($_.FullName)" "$REPO_API/releases/$RELEASE_ID/assets?name=$enc"
+            Write-Host "Uploaded $fn"
+          }
+
+  build-cloud-macos:
+    name: Build Cloud Sidecar (macOS)
+    runs-on: macos-latest
+    env:
+      PYTHON_VERSION: "3.11"
+      RELEASE_TAG: "${{ inputs.tag }}"
+    steps:
+      - name: Show tag
+        run: |
+          echo "Building cloud sidecar for tag ${RELEASE_TAG}"
+
+      - uses: actions/checkout@v4
+        with:
+          ref: ${{ inputs.tag }}
+
+      - name: Install uv
+        run: |
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          echo "$HOME/.local/bin" >> $GITHUB_PATH
+
+      - name: Set up Python
+        run: uv python install ${{ env.PYTHON_VERSION }}
+
+      - name: Install system dependencies
+        run: brew install portaudio
+
+      - name: Build cloud sidecar
+        env:
+          UV_NO_SOURCES: "1"
+        run: |
+          uv venv
+          uv pip install pyinstaller numpy sounddevice fastapi uvicorn websockets pydantic requests pyyaml packaging
+          .venv/bin/pyinstaller local-transcription-cloud.spec
+
+      - name: Package
+        run: |
+          cd dist/local-transcription-backend && zip -r ../../sidecar-macos-aarch64-cloud.zip .
+
+      - name: Upload to release
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          which jq || brew install jq
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+          TAG="${RELEASE_TAG}"
+
+          for i in $(seq 1 30); do
+            RELEASE_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/tags/${TAG}" | jq -r '.id // empty')
+            if [ -n "${RELEASE_ID}" ] && [ "${RELEASE_ID}" != "null" ]; then
+              echo "Found release ${TAG} (ID: ${RELEASE_ID})"
+              break
+            fi
+            echo "Attempt ${i}/30: waiting for release..."
+            sleep 10
+          done
+
+          if [ -z "${RELEASE_ID}" ] || [ "${RELEASE_ID}" = "null" ]; then
+            echo "ERROR: Release not found"; exit 1
+          fi
+
+          for file in sidecar-*-cloud.zip; do
+            filename=$(basename "$file")
+            ASSET_ID=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+              "${REPO_API}/releases/${RELEASE_ID}/assets" | jq -r ".[] | select(.name == \"${filename}\") | .id // empty")
+            [ -n "${ASSET_ID}" ] && curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" "${REPO_API}/releases/${RELEASE_ID}/assets/${ASSET_ID}"
+            curl -s -o /dev/null -w "Upload ${filename}: HTTP %{http_code}\n" -X POST \
+              -H "Authorization: token ${BUILD_TOKEN}" -H "Content-Type: application/octet-stream" \
+              -T "$file" "${REPO_API}/releases/${RELEASE_ID}/assets?name=${filename}"
+          done

.gitea/workflows/build-sidecar-linux.yml

@@ -13,23 +13,15 @@ jobs:
     runs-on: ubuntu-latest
     env:
       PYTHON_VERSION: "3.11"
+      RELEASE_TAG: "${{ inputs.tag }}"
     steps:
-      - name: Determine tag
+      - name: Show tag
-        id: tag
         run: |
-          TAG="${{ inputs.tag }}"
+          echo "Building for tag: ${RELEASE_TAG}"
-          if [ -z "$TAG" ]; then
-            TAG="${{ github.event.inputs.tag }}"
-          fi
-          if [ -z "$TAG" ]; then
-            TAG=$(git ls-remote --tags --sort=-v:refname origin 'refs/tags/sidecar-v*' | head -1 | sed 's|.*refs/tags/||')
-          fi
-          echo "Building for tag: ${TAG}"
-          echo "tag=${TAG}" >> $GITHUB_OUTPUT
 
       - uses: actions/checkout@v4
         with:
-          ref: ${{ steps.tag.outputs.tag }}
+          ref: ${{ inputs.tag }}
 
       - name: Install uv
         run: |
@@ -48,26 +40,16 @@ jobs:
           sudo apt-get update
           sudo apt-get install -y portaudio19-dev
 
-      - name: Build sidecar (CUDA)
-        run: |
-          uv sync --frozen || uv sync
-          uv run pyinstaller local-transcription-headless.spec
-
-      - name: Package sidecar (CUDA)
-        run: |
-          cd dist/local-transcription-backend && zip -r ../../sidecar-linux-x86_64-cuda.zip .
-
       - name: Build sidecar (CPU)
+        env:
+          UV_NO_SOURCES: "1"
         run: |
-          rm -rf dist/local-transcription-backend build/
+          uv sync
-          uv pip install torch torchaudio --index-url https://download.pytorch.org/whl/cpu --force-reinstall
-          # Run pyinstaller directly from venv to prevent uv run from
-          # re-resolving torch back to the CUDA version via pyproject.toml sources
           .venv/bin/pyinstaller local-transcription-headless.spec
 
       - name: Package sidecar (CPU)
         run: |
-          cd dist/local-transcription-backend && zip -r ../../sidecar-linux-x86_64-cpu.zip .
+          cd dist/local-transcription-backend && zip -9 -r ../../sidecar-linux-x86_64-cpu.zip .
 
       - name: Upload to sidecar release
         env:
@@ -75,7 +57,7 @@ jobs:
         run: |
           sudo apt-get install -y jq
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
-          TAG="${{ steps.tag.outputs.tag }}"
+          TAG="${RELEASE_TAG}"
 
           echo "Waiting for sidecar release ${TAG} to be available..."
           for i in $(seq 1 30); do

.gitea/workflows/build-sidecar-macos.yml

@@ -13,23 +13,15 @@ jobs:
     runs-on: macos-latest
     env:
       PYTHON_VERSION: "3.11"
+      RELEASE_TAG: "${{ inputs.tag }}"
     steps:
-      - name: Determine tag
+      - name: Show tag
-        id: tag
         run: |
-          TAG="${{ inputs.tag }}"
+          echo "Building for tag: ${RELEASE_TAG}"
-          if [ -z "$TAG" ]; then
-            TAG="${{ github.event.inputs.tag }}"
-          fi
-          if [ -z "$TAG" ]; then
-            TAG=$(git ls-remote --tags --sort=-v:refname origin 'refs/tags/sidecar-v*' | head -1 | sed 's|.*refs/tags/||')
-          fi
-          echo "Building for tag: ${TAG}"
-          echo "tag=${TAG}" >> $GITHUB_OUTPUT
 
       - uses: actions/checkout@v4
         with:
-          ref: ${{ steps.tag.outputs.tag }}
+          ref: ${{ inputs.tag }}
 
       - name: Install uv
         run: |
@@ -66,7 +58,7 @@ jobs:
         run: |
           which jq || brew install jq
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
-          TAG="${{ steps.tag.outputs.tag }}"
+          TAG="${RELEASE_TAG}"
 
           echo "Waiting for sidecar release ${TAG} to be available..."
           for i in $(seq 1 30); do

.gitea/workflows/build-sidecar-windows.yml

@@ -13,7 +13,7 @@ jobs:
     runs-on: windows-latest
     env:
       PYTHON_VERSION: "3.11"
-      RELEASE_TAG: ${{ inputs.tag }}
+      RELEASE_TAG: "${{ inputs.tag }}"
     steps:
       - name: Show tag
         shell: powershell
@@ -54,29 +54,18 @@ jobs:
             choco install 7zip -y
           }
 
-      - name: Build sidecar (CUDA)
-        shell: powershell
-        run: |
-          uv sync --frozen
-          if ($LASTEXITCODE -ne 0) { uv sync }
-          uv run pyinstaller local-transcription-headless.spec
-
-      - name: Package sidecar (CUDA)
-        shell: powershell
-        run: |
-          7z a -tzip -mx=5 sidecar-windows-x86_64-cuda.zip .\dist\local-transcription-backend\*
-
       - name: Build sidecar (CPU)
         shell: powershell
+        env:
+          UV_NO_SOURCES: "1"
         run: |
-          Remove-Item -Recurse -Force dist\local-transcription-backend, build -ErrorAction SilentlyContinue
+          uv sync
-          uv pip install torch torchaudio --index-url https://download.pytorch.org/whl/cpu --force-reinstall
           .venv\Scripts\pyinstaller.exe local-transcription-headless.spec
 
       - name: Package sidecar (CPU)
         shell: powershell
         run: |
-          7z a -tzip -mx=5 sidecar-windows-x86_64-cpu.zip .\dist\local-transcription-backend\*
+          7z a -tzip -mx=9 sidecar-windows-x86_64-cpu.zip .\dist\local-transcription-backend\*
 
       - name: Upload to sidecar release
         shell: powershell

.gitea/workflows/release.yml

@@ -1,13 +1,40 @@
 name: Release
 
 on:
-  push:
+  workflow_dispatch:
-    branches: [main]
 
 jobs:
+  test:
+    name: Run Tests
+    if: "!contains(github.event.head_commit.message, '[skip ci]')"
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 20
+
+      - name: Install npm deps
+        run: npm ci
+
+      - name: Frontend tests
+        run: npx vitest run
+
+      - name: Install uv
+        run: |
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          echo "$HOME/.local/bin" >> $GITHUB_PATH
+
+      - name: Python tests
+        run: |
+          uv venv .testvenv
+          VIRTUAL_ENV=.testvenv uv pip install pytest httpx pytest-asyncio anyio fastapi pydantic pyyaml uvicorn requests
+          .testvenv/bin/python -m pytest backend/tests/ client/tests/ -v --tb=short
+
   bump-version:
     name: Bump version and tag
-    if: "!contains(github.event.head_commit.message, '[skip ci]')"
+    needs: test
     runs-on: ubuntu-latest
     outputs:
       new_version: ${{ steps.bump.outputs.new_version }}
@@ -82,10 +109,14 @@ jobs:
 
           for workflow in build-app-linux.yml build-app-windows.yml build-app-macos.yml; do
             echo "Dispatching ${workflow} for ${TAG}..."
-            HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST \
+            HTTP_CODE=$(curl -s -w "%{http_code}" -o /tmp/dispatch_resp.txt -X POST \
               -H "Authorization: token ${BUILD_TOKEN}" \
               -H "Content-Type: application/json" \
               -d "{\"ref\": \"main\", \"inputs\": {\"tag\": \"${TAG}\"}}" \
               "${REPO_API}/actions/workflows/${workflow}/dispatches")
             echo "  -> HTTP ${HTTP_CODE}"
+            [ "$HTTP_CODE" != "204" ] && cat /tmp/dispatch_resp.txt && echo ""
           done
+
+      # NOTE: Automatic cleanup disabled -- it races with async builds.
+      # Clean up old releases manually from the Gitea UI when needed.

.gitea/workflows/sidecar-release.yml

@@ -1,19 +1,30 @@
 name: Sidecar Release
 
 on:
-  push:
-    branches: [main]
-    paths:
-      - 'client/**'
-      - 'server/**'
-      - 'backend/**'
-      - 'pyproject.toml'
-      - 'local-transcription-headless.spec'
   workflow_dispatch:
 
 jobs:
+  test:
+    name: Run Tests
+    if: "!contains(github.event.head_commit.message, '[skip ci]')"
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Install uv
+        run: |
+          curl -LsSf https://astral.sh/uv/install.sh | sh
+          echo "$HOME/.local/bin" >> $GITHUB_PATH
+
+      - name: Python tests
+        run: |
+          uv venv .testvenv
+          VIRTUAL_ENV=.testvenv uv pip install pytest httpx pytest-asyncio anyio fastapi pydantic pyyaml uvicorn requests
+          .testvenv/bin/python -m pytest backend/tests/ client/tests/ -v --tb=short
+
   bump-sidecar-version:
     name: Bump sidecar version and tag
+    needs: test
     if: "!contains(github.event.head_commit.message, '[skip ci]')"
     runs-on: ubuntu-latest
     outputs:
@@ -62,8 +73,6 @@ jobs:
           echo "New sidecar version: ${NEW_VERSION}"
 
           sed -i "s/^version = \"${CURRENT}\"/version = \"${NEW_VERSION}\"/" pyproject.toml
-          sed -i "s/__version__ = \"${CURRENT}\"/__version__ = \"${NEW_VERSION}\"/" version.py
-          sed -i "s/__version_info__ = .*/__version_info__ = (${MAJOR}, ${MINOR}, ${NEW_PATCH})/" version.py
 
           echo "version=${NEW_VERSION}" >> $GITHUB_OUTPUT
           echo "tag=sidecar-v${NEW_VERSION}" >> $GITHUB_OUTPUT
@@ -75,7 +84,7 @@ jobs:
         run: |
           NEW_VERSION="${{ steps.bump.outputs.version }}"
           TAG="${{ steps.bump.outputs.tag }}"
-          git add pyproject.toml version.py
+          git add pyproject.toml
           git commit -m "chore: bump sidecar version to ${NEW_VERSION} [skip ci]"
           git tag "${TAG}"
 
@@ -109,7 +118,7 @@ jobs:
           REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
           TAG="${{ steps.bump.outputs.tag }}"
 
-          for workflow in build-sidecar-linux.yml build-sidecar-windows.yml build-sidecar-macos.yml; do
+          for workflow in build-sidecar-linux.yml build-sidecar-windows.yml build-sidecar-macos.yml build-sidecar-cloud.yml; do
             echo "Dispatching ${workflow} for ${TAG}..."
             HTTP_CODE=$(curl -s -o /dev/null -w "%{http_code}" -X POST \
               -H "Authorization: token ${BUILD_TOKEN}" \
@@ -118,3 +127,6 @@ jobs:
               "${REPO_API}/actions/workflows/${workflow}/dispatches")
             echo "  -> HTTP ${HTTP_CODE}"
           done
+
+      # NOTE: Automatic cleanup disabled -- it races with async builds.
+      # Clean up old releases manually from the Gitea UI when needed.

.gitea/workflows/test.yml

@@ -5,6 +5,7 @@ on:
     branches: [main]
   pull_request:
     branches: [main]
+  workflow_dispatch:
 
 jobs:
   python-tests:
@@ -13,12 +14,20 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
-      - name: Install test dependencies
+      - name: Install uv
         run: |
-          pip install --break-system-packages pytest httpx pytest-asyncio anyio fastapi pydantic pyyaml uvicorn requests
+          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: Run pytest
-        run: python3 -m pytest backend/tests/ client/tests/ -v --tb=short
+        run: |
+          uv venv .testvenv
+          VIRTUAL_ENV=.testvenv uv pip install pytest httpx pytest-asyncio anyio fastapi pydantic pyyaml uvicorn requests
+          .testvenv/bin/python -m pytest backend/tests/ client/tests/ -v --tb=short
 
   frontend-tests:
     name: Frontend Tests

backend/app_controller.py

@@ -18,13 +18,18 @@ import sys
 sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
 
 from client.config import Config
-from client.device_utils import DeviceManager
+from client.models import TranscriptionResult
-from client.transcription_engine_realtime import RealtimeTranscriptionEngine, TranscriptionResult
 from client.deepgram_transcription import DeepgramTranscriptionEngine
 from client.server_sync import ServerSyncClient
 from server.web_display import TranscriptionWebServer
 from version import __version__
 
+# Heavy imports (torch, RealtimeSTT, faster-whisper) are deferred so
+# the cloud-only sidecar build can exclude them entirely.
+# Imported lazily in _initialize_engine() when remote.mode == "local".
+RealtimeTranscriptionEngine = None
+DeviceManager = None
+
 
 class AppState:
     """Enum-like class for application states."""
@@ -89,7 +94,24 @@ class AppController:
 
     def __init__(self, config: Optional[Config] = None):
         self.config = config or Config()
-        self.device_manager = DeviceManager()
+
+        # DeviceManager is only needed for local Whisper mode.
+        # Lazy-import to keep the cloud-only sidecar lightweight.
+        global DeviceManager
+        if DeviceManager is None:
+            try:
+                from client.device_utils import DeviceManager as _DM
+                DeviceManager = _DM
+            except ImportError:
+                DeviceManager = None
+
+        self.device_manager = DeviceManager() if DeviceManager else None
+        self.is_cloud_only = DeviceManager is None
+
+        # If this is the cloud-only sidecar and mode is still "local",
+        # auto-switch to "byok" so the engine doesn't try to load Whisper.
+        if self.is_cloud_only and self.config.get('remote.mode', 'local') == 'local':
+            self.config.set('remote.mode', 'byok')
 
         # State
         self._state = AppState.INITIALIZING
@@ -243,15 +265,12 @@ class AppController:
 
     def _initialize_engine(self):
         """Initialize the transcription engine in a background thread."""
-        device_config = self.config.get('transcription.device', 'auto')
-        self.device_manager.set_device(device_config)
-
         audio_device_str = self.config.get('audio.input_device', 'default')
         audio_device = None if audio_device_str == 'default' else int(audio_device_str)
 
         model = self.config.get('transcription.model', 'base.en')
         language = self.config.get('transcription.language', 'en')
-        device = self.device_manager.get_device_for_whisper()
+        device_config = self.config.get('transcription.device', 'auto')
         compute_type = self.config.get('transcription.compute_type', 'default')
 
         self.current_model_size = model
@@ -284,6 +303,27 @@ class AppController:
             self.transcription_engine.set_error_callback(self._on_remote_error)
             self.transcription_engine.set_credits_low_callback(self._on_credits_low)
         else:
+            # Lazy-import heavy local transcription dependencies
+            global RealtimeTranscriptionEngine
+            if RealtimeTranscriptionEngine is None:
+                try:
+                    from client.transcription_engine_realtime import RealtimeTranscriptionEngine as _RTE
+                    RealtimeTranscriptionEngine = _RTE
+                except ImportError:
+                    # Cloud-only sidecar -- local engine not available
+                    self._set_state(
+                        AppState.ERROR,
+                        "Local transcription not available in this build. "
+                        "Please switch to Cloud (Deepgram) mode in Settings."
+                    )
+                    return
+
+            if self.device_manager:
+                self.device_manager.set_device(device_config)
+                device = self.device_manager.get_device_for_whisper()
+            else:
+                device = "cpu"
+
             self.transcription_engine = RealtimeTranscriptionEngine(
                 model=model,
                 device=device,
@@ -333,7 +373,15 @@ class AppController:
 
             self._set_state(AppState.READY, f"Ready | Device: {device_display}")
         else:
-            self._set_state(AppState.ERROR, message)
+            # Cloud sidecar with no API key -- show helpful setup message
+            # instead of a scary error. The user needs to enter their key.
+            if self.is_cloud_only:
+                self._set_state(
+                    AppState.READY,
+                    "Setup needed: Open Settings > Remote Transcription > enter your Deepgram API key"
+                )
+            else:
+                self._set_state(AppState.ERROR, message)
 
     # ── Transcription Control ──────────────────────────────────────
 
@@ -577,12 +625,18 @@ class AppController:
             if self.config.get('server_sync.enabled', False):
                 self._start_server_sync()
 
-        # Check if model/device changed
+        # Check if model/device/remote mode changed -- any of these require
+        # a full engine reload since they change which engine class is used
         new_model = self.config.get('transcription.model', 'base.en')
         new_device = self.config.get('transcription.device', 'auto')
+        new_remote_mode = self.config.get('remote.mode', 'local')
+        current_remote_mode = 'local'
+        if self.transcription_engine:
+            current_remote_mode = getattr(self.transcription_engine, 'mode', 'local')
         engine_reload_needed = (
             self.current_model_size != new_model
             or self.current_device_config != new_device
+            or current_remote_mode != new_remote_mode
         )
 
         if engine_reload_needed:
@@ -596,7 +650,7 @@ class AppController:
         host = self.config.get('web_server.host', '127.0.0.1')
         port = self.actual_web_port or self.config.get('web_server.port', 8080)
 
-        device_info = self.device_manager.get_device_info()
+        device_info = self.device_manager.get_device_info() if self.device_manager else []
 
         remote_mode = self.config.get('remote.mode', 'local')
         if remote_mode in ('managed', 'byok') and self.transcription_engine:
@@ -640,10 +694,13 @@ class AppController:
 
     def get_compute_devices(self) -> list[dict]:
         """List available compute devices."""
-        device_info = self.device_manager.get_device_info()
         devices = [{"id": "auto", "name": "Auto-detect"}]
-        for dev_id, dev_name in device_info:
+        if self.device_manager:
-            devices.append({"id": dev_id, "name": dev_name})
+            device_info = self.device_manager.get_device_info()
+            for dev_id, dev_name in device_info:
+                devices.append({"id": dev_id, "name": dev_name})
+        else:
+            devices.append({"id": "cloud", "name": "Cloud (Deepgram)"})
         return devices
 
     # ── Update Checking ────────────────────────────────────────────

backend/tests/test_api_server.py

@@ -79,7 +79,7 @@ async def test_start_when_not_ready(api_client, controller):
 
 @pytest.mark.asyncio
 async def test_clear(api_client, controller):
-    from client.transcription_engine_realtime import TranscriptionResult
+    from client.models import TranscriptionResult
     from datetime import datetime
 
     controller.transcriptions = [

backend/tests/test_app_controller.py

@@ -72,7 +72,7 @@ def test_double_start_rejected(controller):
 
 def test_clear_transcriptions(controller):
     """clear_transcriptions should empty the list and return the count."""
-    from client.transcription_engine_realtime import TranscriptionResult
+    from client.models import TranscriptionResult
 
     controller.transcriptions = [
         TranscriptionResult(text="Hello", is_final=True, timestamp=datetime.now(), user_name="Alice"),
@@ -85,7 +85,7 @@ def test_clear_transcriptions(controller):
 
 def test_get_transcriptions_text_with_timestamps(controller):
     """get_transcriptions_text should include [HH:MM:SS] prefixes when requested."""
-    from client.transcription_engine_realtime import TranscriptionResult
+    from client.models import TranscriptionResult
 
     ts = datetime(2025, 1, 15, 10, 30, 45)
     controller.transcriptions = [
@@ -141,7 +141,7 @@ def test_apply_settings_no_reload_when_same(controller):
 
 def test_on_final_transcription_callback_fires(controller):
     """_on_final_transcription should append and invoke on_transcription callback."""
-    from client.transcription_engine_realtime import TranscriptionResult
+    from client.models import TranscriptionResult
 
     received = []
     controller.on_transcription = lambda data: received.append(data)
@@ -166,7 +166,7 @@ def test_on_final_transcription_callback_fires(controller):
 
 def test_on_final_transcription_ignored_when_not_transcribing(controller):
     """If the controller is not in transcribing state the callback should be a no-op."""
-    from client.transcription_engine_realtime import TranscriptionResult
+    from client.models import TranscriptionResult
 
     controller.is_transcribing = False
 

client/deepgram_transcription.py

@@ -17,7 +17,7 @@ from datetime import datetime
 from queue import Queue, Empty
 from typing import Optional, Callable
 
-from client.transcription_engine_realtime import TranscriptionResult
+from client.models import TranscriptionResult
 
 logger = logging.getLogger(__name__)
 
@@ -67,7 +67,7 @@ class DeepgramTranscriptionEngine:
         # Audio parameters
         self.sample_rate: int = 16000
         self.channels: int = 1
-        self.blocksize: int = 4096
+        self.blocksize: int = 1024  # ~64ms chunks for lower latency streaming
 
         # Callbacks
         self.realtime_callback: Optional[Callable[[TranscriptionResult], None]] = None
@@ -314,6 +314,8 @@ class DeepgramTranscriptionEngine:
             f"model={self.deepgram_model}"
             f"&language={self.language}"
             "&interim_results=true"
+            "&punctuate=true"
+            "&smart_format=true"
             "&encoding=linear16"
             f"&sample_rate={self.sample_rate}"
             f"&channels={self.channels}"
@@ -370,10 +372,16 @@ class DeepgramTranscriptionEngine:
 
     async def _send_loop(self):
         """Drain the audio queue and push raw PCM bytes over the WebSocket."""
+        loop = asyncio.get_event_loop()
         while not self._stop_event.is_set():
             try:
-                pcm_bytes = self._audio_queue.get(timeout=0.1)
+                # Use run_in_executor to avoid blocking the async event loop
-            except Empty:
+                # (which would stall the receive loop and delay transcriptions)
+                pcm_bytes = await asyncio.wait_for(
+                    loop.run_in_executor(None, lambda: self._audio_queue.get(timeout=0.5)),
+                    timeout=1.0,
+                )
+            except (Empty, asyncio.TimeoutError):
                 continue
 
             try:

client/models.py

@@ -0,0 +1,29 @@
+"""Shared data models used across transcription engines."""
+
+from datetime import datetime
+
+
+class TranscriptionResult:
+    """Represents a transcription result."""
+
+    def __init__(self, text: str, is_final: bool, timestamp: datetime, user_name: str = ""):
+        """
+        Initialize transcription result.
+
+        Args:
+            text: Transcribed text
+            is_final: Whether this is a final transcription or realtime preview
+            timestamp: Timestamp of transcription
+            user_name: Name of the user/speaker
+        """
+        self.text = text.strip()
+        self.is_final = is_final
+        self.timestamp = timestamp
+        self.user_name = user_name
+
+    def __repr__(self) -> str:
+        time_str = self.timestamp.strftime("%H:%M:%S")
+        prefix = "[FINAL]" if self.is_final else "[PREVIEW]"
+        if self.user_name and self.user_name.strip():
+            return f"{prefix} [{time_str}] {self.user_name}: {self.text}"
+        return f"{prefix} [{time_str}] {self.text}"

client/transcription_engine_realtime.py

@@ -8,30 +8,8 @@ from threading import Lock
 import logging
 
 
-class TranscriptionResult:
+# Re-export TranscriptionResult from the shared models module for backward compatibility
-    """Represents a transcription result."""
+from client.models import TranscriptionResult  # noqa: F401
-
-    def __init__(self, text: str, is_final: bool, timestamp: datetime, user_name: str = ""):
-        """
-        Initialize transcription result.
-
-        Args:
-            text: Transcribed text
-            is_final: Whether this is a final transcription or realtime preview
-            timestamp: Timestamp of transcription
-            user_name: Name of the user/speaker
-        """
-        self.text = text.strip()
-        self.is_final = is_final
-        self.timestamp = timestamp
-        self.user_name = user_name
-
-    def __repr__(self) -> str:
-        time_str = self.timestamp.strftime("%H:%M:%S")
-        prefix = "[FINAL]" if self.is_final else "[PREVIEW]"
-        if self.user_name and self.user_name.strip():
-            return f"{prefix} [{time_str}] {self.user_name}: {self.text}"
-        return f"{prefix} [{time_str}] {self.text}"
 
     def to_dict(self) -> dict:
         """Convert to dictionary."""

local-transcription-cloud.spec

@@ -0,0 +1,152 @@
+# -*- mode: python ; coding: utf-8 -*-
+"""PyInstaller spec file for cloud-only Local Transcription backend.
+
+This builds a lightweight sidecar (~50MB) that only supports Deepgram
+cloud transcription (managed + BYOK). No local Whisper models, no
+PyTorch, no CUDA -- just audio capture and WebSocket streaming.
+"""
+
+import sys
+import os
+
+block_cipher = None
+is_windows = sys.platform == 'win32'
+
+from PyInstaller.utils.hooks import collect_submodules, collect_data_files
+
+# Data files
+datas = [
+    ('config/default_config.yaml', 'config'),
+]
+
+# Hidden imports -- only lightweight deps needed for Deepgram streaming
+hiddenimports = [
+    'sounddevice',
+    'numpy',
+    # FastAPI and dependencies
+    'fastapi',
+    'fastapi.routing',
+    'fastapi.responses',
+    'starlette',
+    'starlette.applications',
+    'starlette.routing',
+    'starlette.responses',
+    'starlette.websockets',
+    'starlette.middleware',
+    'starlette.middleware.cors',
+    'pydantic',
+    'pydantic.fields',
+    'pydantic.main',
+    'anyio',
+    'anyio._backends',
+    'anyio._backends._asyncio',
+    'sniffio',
+    # Uvicorn
+    'uvicorn',
+    'uvicorn.logging',
+    'uvicorn.loops',
+    'uvicorn.loops.auto',
+    'uvicorn.protocols',
+    'uvicorn.protocols.http',
+    'uvicorn.protocols.http.auto',
+    'uvicorn.protocols.http.h11_impl',
+    'uvicorn.protocols.websockets',
+    'uvicorn.protocols.websockets.auto',
+    'uvicorn.protocols.websockets.wsproto_impl',
+    'uvicorn.lifespan',
+    'uvicorn.lifespan.on',
+    'h11',
+    'websockets',
+    'websockets.legacy',
+    'websockets.legacy.server',
+    # HTTP client
+    'requests',
+    'urllib3',
+    'certifi',
+    'charset_normalizer',
+]
+
+# Collect submodules for key packages
+print("Collecting submodules for cloud backend packages...")
+for package in ['fastapi', 'starlette', 'pydantic', 'pydantic_core', 'anyio', 'uvicorn', 'websockets', 'h11']:
+    try:
+        submodules = collect_submodules(package)
+        hiddenimports += submodules
+        print(f"  + Collected {len(submodules)} submodules from {package}")
+    except Exception as e:
+        print(f"  - Warning: Could not collect {package}: {e}")
+
+# Collect data files
+for package in ['fastapi', 'starlette', 'pydantic', 'uvicorn']:
+    try:
+        data_files = collect_data_files(package)
+        if data_files:
+            datas += data_files
+    except Exception:
+        pass
+
+# Pydantic critical deps
+hiddenimports += [
+    'colorsys', 'decimal', 'json', 'ipaddress', 'pathlib', 'uuid',
+    'email.message', 'typing_extensions',
+]
+
+a = Analysis(
+    ['backend/main_headless.py'],
+    pathex=[],
+    binaries=[],
+    datas=datas,
+    hiddenimports=hiddenimports,
+    hookspath=['hooks'],
+    hooksconfig={},
+    runtime_hooks=[],
+    excludes=[
+        # Exclude all heavy ML/local transcription deps
+        'torch', 'torchaudio', 'torchvision',
+        'faster_whisper', 'ctranslate2',
+        'RealtimeSTT', 'webrtcvad', 'webrtcvad_wheels',
+        'silero_vad', 'onnxruntime',
+        'openwakeword', 'pvporcupine', 'pyaudio',
+        'noisereduce', 'scipy',
+        # Exclude GUI frameworks
+        'PySide6', 'PyQt5', 'PyQt6', 'tkinter',
+        # Exclude other unnecessary heavy packages
+        'matplotlib', 'PIL', 'cv2',
+    ],
+    win_no_prefer_redirects=False,
+    win_private_assemblies=False,
+    cipher=block_cipher,
+    noarchive=False,
+)
+
+pyz = PYZ(a.pure, a.zipped_data, cipher=block_cipher)
+
+exe = EXE(
+    pyz,
+    a.scripts,
+    [],
+    exclude_binaries=True,
+    name='local-transcription-backend',
+    debug=False,
+    bootloader_ignore_signals=False,
+    strip=False,
+    upx=True,
+    console=True,
+    disable_windowed_traceback=False,
+    argv_emulation=False,
+    target_arch=None,
+    codesign_identity=None,
+    entitlements_file=None,
+    icon='LocalTranscription.ico' if is_windows else None,
+)
+
+coll = COLLECT(
+    exe,
+    a.binaries,
+    a.zipfiles,
+    a.datas,
+    strip=False,
+    upx=True,
+    upx_exclude=[],
+    name='local-transcription-backend',
+)

package.json

@@ -1,7 +1,7 @@
 {
   "name": "local-transcription",
   "private": true,
-  "version": "1.4.18",
+  "version": "2.0.8",
   "type": "module",
   "scripts": {
     "dev": "vite dev",

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "local-transcription"
-version = "1.0.3"
+version = "1.0.7"
 description = "A standalone desktop application for real-time speech-to-text transcription using Whisper models"
 readme = "README.md"
 requires-python = ">=3.9"

src-tauri/Cargo.lock

@@ -1881,7 +1881,7 @@ checksum = "92daf443525c4cce67b150400bc2316076100ce0b3686209eb8cf3c31612e6f0"
 
 [[package]]
 name = "local-transcription"
-version = "1.4.16"
+version = "2.0.3"
 dependencies = [
  "bytes",
  "chrono",

src-tauri/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "local-transcription"
-version = "1.4.18"
+version = "2.0.8"
 description = "Real-time speech-to-text transcription for streamers"
 authors = ["Local Transcription Contributors"]
 edition = "2021"

src-tauri/gen/schemas/capabilities.json

@@ -1 +1 @@
-{}
+{"default":{"identifier":"default","description":"Default permissions for the main window","local":true,"windows":["main"],"permissions":["core:default","core:event:default","core:event:allow-listen","core:event:allow-emit","shell:default","dialog:default","process:default"]}}

src-tauri/src/lib.rs

@@ -29,9 +29,9 @@ pub fn run() {
         .plugin(tauri_plugin_shell::init())
         .plugin(tauri_plugin_dialog::init())
         .plugin(tauri_plugin_process::init())
-        .manage(sidecar::ManagedSidecar(Mutex::new(
+        .manage(sidecar::ManagedSidecar(std::sync::Arc::new(Mutex::new(
             sidecar::SidecarManager::new(),
-        )))
+        ))))
         .setup(|app| {
             let resource_dir = app
                 .path()
@@ -68,6 +68,7 @@ pub fn run() {
             sidecar::get_sidecar_port,
             sidecar::start_sidecar,
             sidecar::stop_sidecar,
+            sidecar::reset_sidecar,
             write_log,
         ])
         .run(tauri::generate_context!())

src-tauri/src/sidecar/mod.rs

@@ -54,7 +54,8 @@ fn read_installed_version() -> Option<String> {
 }
 
 fn sidecar_dir_for_version(version: &str) -> PathBuf {
-    data_dir().join(format!("sidecar-{version}"))
+    // version is the full tag name, e.g. "sidecar-v1.0.3" -- use it directly
+    data_dir().join(version)
 }
 
 fn binary_path_for_version(version: &str) -> PathBuf {
@@ -371,12 +372,12 @@ fn extract_zip(zip_path: &std::path::Path, dest: &std::path::Path) -> Result<(),
 
 fn cleanup_old_versions(current_version: &str) {
     let data = data_dir();
-    let current_dir_name = format!("sidecar-{current_version}");
+    // current_version is already the full tag, e.g. "sidecar-v1.0.3"
     if let Ok(entries) = std::fs::read_dir(data) {
         for entry in entries.flatten() {
             let name = entry.file_name().to_string_lossy().to_string();
-            if name.starts_with("sidecar-v")      // e.g. sidecar-v1.0.1
+            if name.starts_with("sidecar-")
-                && name != current_dir_name
+                && name != current_version
                 && entry.path().is_dir()
             {
                 let _ = std::fs::remove_dir_all(entry.path());
@@ -433,6 +434,20 @@ impl SidecarManager {
                 .ok_or_else(|| "Sidecar running but port unknown".into());
         }
 
+        // Clear stale PID lock from a previous crash so the sidecar can start.
+        // The Python InstanceLock writes to ~/.local-transcription/app.lock
+        if let Ok(home) = std::env::var("USERPROFILE")
+            .or_else(|_| std::env::var("HOME"))
+        {
+            let lock_file = PathBuf::from(home)
+                .join(".local-transcription")
+                .join("app.lock");
+            if lock_file.exists() {
+                eprintln!("[sidecar] Removing stale lock file: {}", lock_file.display());
+                let _ = std::fs::remove_file(&lock_file);
+            }
+        }
+
         let is_dev = cfg!(debug_assertions)
             || std::env::var("LOCAL_TRANSCRIPTION_DEV")
                 .map(|v| v == "1")
@@ -463,11 +478,63 @@ impl SidecarManager {
             .take()
             .ok_or("Failed to capture sidecar stdout")?;
 
-        let port = Self::wait_for_ready(stdout)?;
+        // Capture stderr in a background thread so we can log it
+        let stderr = child
+            .stderr
+            .take()
+            .ok_or("Failed to capture sidecar stderr")?;
 
-        self.child = Some(child);
+        let log_dir = DIRS.get().map(|d| d.data_dir.clone());
-        self.port = Some(port);
+        std::thread::spawn(move || {
-        Ok(port)
+            use std::io::BufRead;
+            let reader = std::io::BufReader::new(stderr);
+            let mut log_file = log_dir.and_then(|d| {
+                std::fs::OpenOptions::new()
+                    .create(true)
+                    .append(true)
+                    .open(d.join("sidecar.log"))
+                    .ok()
+            });
+            for line in reader.lines() {
+                if let Ok(line) = line {
+                    eprintln!("[sidecar-stderr] {}", line);
+                    if let Some(ref mut f) = log_file {
+                        use std::io::Write;
+                        let _ = writeln!(f, "{}", line);
+                    }
+                }
+            }
+        });
+
+        match Self::wait_for_ready(stdout) {
+            Ok(port) => {
+                self.child = Some(child);
+                self.port = Some(port);
+                Ok(port)
+            }
+            Err(e) => {
+                // Kill the child if ready failed
+                let _ = child.kill();
+                let _ = child.wait();
+
+                // Read the sidecar.log for context
+                let log_hint = DIRS
+                    .get()
+                    .and_then(|d| std::fs::read_to_string(d.data_dir.join("sidecar.log")).ok())
+                    .and_then(|s| {
+                        let lines: Vec<&str> = s.lines().collect();
+                        let tail: Vec<&str> = lines.iter().rev().take(10).rev().cloned().collect();
+                        if tail.is_empty() { None } else { Some(tail.join("\n")) }
+                    })
+                    .unwrap_or_default();
+
+                if log_hint.is_empty() {
+                    Err(e)
+                } else {
+                    Err(format!("{e}\n\nSidecar stderr (last 10 lines):\n{log_hint}"))
+                }
+            }
+        }
     }
 
     /// Stop the sidecar process if running.
@@ -487,20 +554,30 @@ impl SidecarManager {
     // -- private helpers -------------------------------------------------------
 
     fn build_dev_command(&self) -> Result<std::process::Command, String> {
-        let mut cmd = std::process::Command::new("python");
+        // Use `uv run` to ensure we use the project's venv, not system Python
-        cmd.args(["-m", "backend.main_headless"]);
+        let mut cmd = std::process::Command::new("uv");
+        cmd.args(["run", "python", "-u", "-m", "backend.main_headless"]);
 
-        // Try to find the project root (parent of src-tauri)
+        // Find the project root: try CARGO_MANIFEST_DIR first (set at compile time),
-        if let Some(dirs) = DIRS.get() {
+        // then fall back to resource_dir parent chain
-            let project_root = dirs
+        let manifest_dir = option_env!("CARGO_MANIFEST_DIR").map(std::path::PathBuf::from);
-                .resource_dir
+        let project_root = manifest_dir
-                .parent() // src-tauri
+            .as_ref()
-                .and_then(|p| p.parent()); // project root
+            .and_then(|d| d.parent()) // src-tauri -> project root
-            if let Some(root) = project_root {
+            .or_else(|| {
-                cmd.current_dir(root);
+                DIRS.get()
-            }
+                    .and_then(|d| d.resource_dir.parent())
+                    .and_then(|p| p.parent())
+            });
+
+        if let Some(root) = project_root {
+            eprintln!("[sidecar] Dev mode: working dir = {}", root.display());
+            cmd.current_dir(root);
+        } else {
+            eprintln!("[sidecar] Dev mode: WARNING - could not determine project root");
         }
 
+        cmd.env("PYTHONUNBUFFERED", "1");
         Ok(cmd)
     }
 
@@ -516,27 +593,51 @@ impl SidecarManager {
             bin.parent()
                 .ok_or("Cannot determine sidecar parent dir")?,
         );
+        // Force unbuffered stdout so the ready event is sent immediately.
+        // PyInstaller frozen executables buffer stdout when piped.
+        cmd.env("PYTHONUNBUFFERED", "1");
         Ok(cmd)
     }
 
     fn wait_for_ready(stdout: std::process::ChildStdout) -> Result<u16, String> {
-        let reader = std::io::BufReader::new(stdout);
+        use std::sync::mpsc;
-        let timeout = std::time::Duration::from_secs(120);
-        let start = std::time::Instant::now();
 
-        for line in reader.lines() {
+        let timeout = std::time::Duration::from_secs(120);
-            if start.elapsed() > timeout {
+
-                return Err("Timed out waiting for sidecar ready event".into());
+        // Read stdout in a background thread so we can enforce a real timeout.
-            }
+        // BufReader::lines() blocks indefinitely if no data arrives.
-            let line = line.map_err(|e| format!("IO error reading stdout: {e}"))?;
+        let (tx, rx) = mpsc::channel();
-            if let Ok(evt) = serde_json::from_str::<ReadyEvent>(&line) {
+
-                if evt.event == "ready" {
+        std::thread::spawn(move || {
-                    return Ok(evt.port);
+            let reader = std::io::BufReader::new(stdout);
+            for line in reader.lines() {
+                match line {
+                    Ok(line) => {
+                        eprintln!("[sidecar-stdout] {}", line);
+                        if let Ok(evt) = serde_json::from_str::<ReadyEvent>(&line) {
+                            if evt.event == "ready" {
+                                let _ = tx.send(Ok(evt.port));
+                                return;
+                            }
+                        }
+                    }
+                    Err(e) => {
+                        let _ = tx.send(Err(format!("IO error reading stdout: {e}")));
+                        return;
+                    }
                 }
             }
-            // Ignore other lines (e.g. log output)
+            let _ = tx.send(Err(
-        }
+                "Sidecar process exited before sending ready event".into(),
-        Err("Sidecar process exited before sending ready event".into())
+            ));
+        });
+
+        rx.recv_timeout(timeout).unwrap_or_else(|_| {
+            Err(format!(
+                "Timed out after {}s waiting for sidecar ready event",
+                timeout.as_secs()
+            ))
+        })
     }
 }
 
@@ -545,7 +646,8 @@ impl SidecarManager {
 // ---------------------------------------------------------------------------
 
 /// Wrapper so we can store `SidecarManager` in Tauri's managed state.
-pub struct ManagedSidecar(pub Mutex<SidecarManager>);
+/// Uses Arc so it can be cloned into background threads for async commands.
+pub struct ManagedSidecar(pub std::sync::Arc<Mutex<SidecarManager>>);
 
 #[tauri::command]
 pub fn get_sidecar_port(state: tauri::State<'_, ManagedSidecar>) -> Result<Option<u16>, String> {
@@ -561,12 +663,16 @@ pub fn get_sidecar_port(state: tauri::State<'_, ManagedSidecar>) -> Result<Optio
 }
 
 #[tauri::command]
-pub fn start_sidecar(state: tauri::State<'_, ManagedSidecar>) -> Result<u16, String> {
+pub async fn start_sidecar(state: tauri::State<'_, ManagedSidecar>) -> Result<u16, String> {
-    let mut mgr = state
+    let mgr = state.0.clone();
-        .0
+    // Run blocking sidecar launch in a background thread so it doesn't
-        .lock()
+    // freeze the Tauri UI while waiting for the ready event (up to 120s).
-        .map_err(|e| format!("Lock error: {e}"))?;
+    tokio::task::spawn_blocking(move || {
-    mgr.ensure_running()
+        let mut mgr = mgr.lock().map_err(|e| format!("Lock error: {e}"))?;
+        mgr.ensure_running()
+    })
+    .await
+    .map_err(|e| format!("Task join error: {e}"))?
 }
 
 #[tauri::command]
@@ -579,6 +685,42 @@ pub fn stop_sidecar(state: tauri::State<'_, ManagedSidecar>) -> Result<(), Strin
     Ok(())
 }
 
+/// Stop the running sidecar, delete its files and version marker.
+/// The next app launch will show the sidecar download prompt.
+#[tauri::command]
+pub fn reset_sidecar(state: tauri::State<'_, ManagedSidecar>) -> Result<(), String> {
+    // Stop the running sidecar first
+    {
+        let mut mgr = state
+            .0
+            .lock()
+            .map_err(|e| format!("Lock error: {e}"))?;
+        mgr.stop();
+    }
+
+    let data = data_dir();
+
+    // Delete the version file so check_sidecar returns false
+    let vf = version_file();
+    if vf.exists() {
+        std::fs::remove_file(&vf)
+            .map_err(|e| format!("Failed to delete version file: {e}"))?;
+    }
+
+    // Delete all sidecar directories
+    if let Ok(entries) = std::fs::read_dir(&data) {
+        for entry in entries.flatten() {
+            let name = entry.file_name().to_string_lossy().to_string();
+            if name.starts_with("sidecar-") && entry.path().is_dir() {
+                eprintln!("[sidecar] Removing {}", entry.path().display());
+                let _ = std::fs::remove_dir_all(entry.path());
+            }
+        }
+    }
+
+    Ok(())
+}
+
 // ---------------------------------------------------------------------------
 // Tests
 // ---------------------------------------------------------------------------
@@ -739,7 +881,7 @@ mod tests {
     fn sidecar_dir_for_version_contains_version() {
         let data = ensure_dirs_initialised();
         let dir = sidecar_dir_for_version("sidecar-v1.2.3");
-        assert_eq!(dir, data.join("sidecar-sidecar-v1.2.3"));
+        assert_eq!(dir, data.join("sidecar-v1.2.3"));
     }
 
     #[test]
@@ -784,9 +926,8 @@ mod tests {
             std::fs::create_dir_all(data.join(d)).unwrap();
         }
 
-        // cleanup_old_versions builds `current_dir_name = "sidecar-{version}"`.
+        // current_version is the full tag, e.g. "sidecar-v1.0.2"
-        // Passing "v1.0.2" produces "sidecar-v1.0.2" which matches our dir name.
+        cleanup_old_versions("sidecar-v1.0.2");
-        cleanup_old_versions("v1.0.2");
 
         assert!(
             !data.join("sidecar-v1.0.0").exists(),

src-tauri/tauri.conf.json

@@ -1,6 +1,6 @@
 {
   "productName": "Local Transcription",
-  "version": "1.4.18",
+  "version": "2.0.8",
   "identifier": "net.anhonesthost.local-transcription",
   "build": {
     "frontendDist": "../dist",

src/App.svelte

@@ -9,11 +9,12 @@
   import { backendStore } from "$lib/stores/backend";
   import { configStore } from "$lib/stores/config";
 
-  type SidecarState = "checking" | "needs_setup" | "starting" | "connected";
+  type SidecarState = "checking" | "needs_setup" | "update_available" | "starting" | "connected";
 
   let showSettings = $state(false);
   let sidecarState = $state<SidecarState>("checking");
   let debugLog = $state("");
+  let availableUpdate = $state("");
 
   let obsDisplayUrl = $derived(backendStore.obsUrl);
   let syncDisplayUrl = $derived(backendStore.syncUrl);
@@ -53,6 +54,20 @@
         return;
       }
 
+      // Check for sidecar updates before launching
+      try {
+        log("Checking for sidecar updates...");
+        const update = await invoke<string | null>("check_sidecar_update");
+        if (update) {
+          log(`Sidecar update available: ${update}`);
+          availableUpdate = update;
+          sidecarState = "update_available";
+          return;
+        }
+      } catch (err) {
+        log(`Update check failed (non-fatal): ${err}`);
+      }
+
       await launchSidecar();
     } catch (err) {
       // Not running in Tauri (browser dev mode) - skip sidecar check
@@ -118,6 +133,26 @@
 {:else if sidecarState === "needs_setup"}
   <SidecarSetup onComplete={onSidecarReady} />
 
+{:else if sidecarState === "update_available"}
+  <div class="connecting-overlay" style="background:#1e1e1e;color:#e0e0e0;display:flex;align-items:center;justify-content:center;height:100%;width:100%;">
+    <div class="connecting-content" style="text-align:center;max-width:400px;">
+      <h2 style="font-size:20px;margin:0 0 12px;">Sidecar Update Available</h2>
+      <p style="color:#a0a0a0;font-size:14px;margin:0 0 20px;">
+        A new version of the transcription engine is available ({availableUpdate}).
+      </p>
+      <div style="display:flex;gap:10px;justify-content:center;">
+        <button
+          style="padding:8px 20px;border:1px solid #555;border-radius:6px;background:transparent;color:#e0e0e0;cursor:pointer;"
+          onclick={() => launchSidecar()}
+        >Skip</button>
+        <button
+          style="padding:8px 20px;border:none;border-radius:6px;background:#4CAF50;color:white;cursor:pointer;font-weight:500;"
+          onclick={() => { sidecarState = "needs_setup"; }}
+        >Update Now</button>
+      </div>
+    </div>
+  </div>
+
 {:else if !isConnected}
   <div class="connecting-overlay" style="background:#1e1e1e;color:#e0e0e0;display:flex;align-items:center;justify-content:center;height:100%;width:100%;">
     <div class="connecting-content" style="text-align:center;">

src/lib/components/Controls.svelte

@@ -17,6 +17,9 @@
       } else {
         await backendStore.apiPost("/api/start");
       }
+      // Poll status to update UI immediately instead of waiting
+      // for WebSocket broadcast (which can be delayed or missed)
+      await backendStore.pollStatus();
     } catch (err) {
       console.error("Failed to toggle transcription:", err);
     } finally {

src/lib/components/Settings.svelte

@@ -27,6 +27,10 @@
   let showTimestamps = $state(true);
   let fadeSeconds = $state(10);
   let maxLines = $state(100);
+  let fontSource = $state("System Font");
+  let fontFamily = $state("Courier");
+  let websafeFont = $state("Arial");
+  let googleFont = $state("Roboto");
   let fontSize = $state(12);
   let userColor = $state("#4CAF50");
   let textColor = $state("#FFFFFF");
@@ -99,6 +103,10 @@
     showTimestamps = cfg.display.show_timestamps;
     fadeSeconds = cfg.display.fade_after_seconds;
     maxLines = cfg.display.max_lines;
+    fontSource = cfg.display.font_source ?? "System Font";
+    fontFamily = cfg.display.font_family ?? "Courier";
+    websafeFont = cfg.display.websafe_font ?? "Arial";
+    googleFont = cfg.display.google_font ?? "Roboto";
     fontSize = cfg.display.font_size;
     userColor = cfg.display.user_color;
     textColor = cfg.display.text_color;
@@ -174,6 +182,10 @@
         show_timestamps: showTimestamps,
         fade_after_seconds: fadeSeconds,
         max_lines: maxLines,
+        font_source: fontSource,
+        font_family: fontFamily,
+        websafe_font: websafeFont,
+        google_font: googleFont,
         font_size: fontSize,
         user_color: userColor,
         text_color: textColor,
@@ -220,6 +232,18 @@
     }
   }
 
+  async function handleChangeSidecar() {
+    try {
+      const { invoke } = await import("@tauri-apps/api/core");
+      await invoke("reset_sidecar");
+      // Force a page reload which will re-trigger the setup flow
+      window.location.reload();
+    } catch (err) {
+      console.error("Failed to reset sidecar:", err);
+      saveMessage = `Error: ${err}`;
+    }
+  }
+
   async function handleManagedLogin() {
     try {
       await backendStore.apiPost("/api/login", {
@@ -485,6 +509,95 @@
             bind:value={maxLines}
           />
         </div>
+        <div class="field">
+          <label for="font-source">Font Source</label>
+          <select id="font-source" bind:value={fontSource}>
+            <option value="System Font">System Font</option>
+            <option value="Web-Safe">Web-Safe</option>
+            <option value="Google Font">Google Font</option>
+          </select>
+        </div>
+        {#if fontSource === "System Font"}
+          <div class="field">
+            <label for="font-family">System Font Family</label>
+            <input id="font-family" type="text" bind:value={fontFamily} placeholder="Courier" />
+          </div>
+        {/if}
+        {#if fontSource === "Web-Safe"}
+          <div class="field">
+            <label for="websafe-font">Web-Safe Font</label>
+            <select id="websafe-font" bind:value={websafeFont}>
+              <option value="Arial">Arial</option>
+              <option value="Arial Black">Arial Black</option>
+              <option value="Comic Sans MS">Comic Sans MS</option>
+              <option value="Courier New">Courier New</option>
+              <option value="Georgia">Georgia</option>
+              <option value="Impact">Impact</option>
+              <option value="Lucida Console">Lucida Console</option>
+              <option value="Lucida Sans Unicode">Lucida Sans Unicode</option>
+              <option value="Palatino Linotype">Palatino Linotype</option>
+              <option value="Tahoma">Tahoma</option>
+              <option value="Times New Roman">Times New Roman</option>
+              <option value="Trebuchet MS">Trebuchet MS</option>
+              <option value="Verdana">Verdana</option>
+            </select>
+          </div>
+        {/if}
+        {#if fontSource === "Google Font"}
+          <div class="field">
+            <label for="google-font">Google Font</label>
+            <select id="google-font" bind:value={googleFont}>
+              <optgroup label="Sans Serif">
+                <option value="Roboto">Roboto</option>
+                <option value="Open Sans">Open Sans</option>
+                <option value="Lato">Lato</option>
+                <option value="Montserrat">Montserrat</option>
+                <option value="Poppins">Poppins</option>
+                <option value="Nunito">Nunito</option>
+                <option value="Raleway">Raleway</option>
+                <option value="Ubuntu">Ubuntu</option>
+                <option value="Rubik">Rubik</option>
+                <option value="Work Sans">Work Sans</option>
+                <option value="Inter">Inter</option>
+                <option value="Outfit">Outfit</option>
+                <option value="Quicksand">Quicksand</option>
+                <option value="Comfortaa">Comfortaa</option>
+                <option value="Varela Round">Varela Round</option>
+              </optgroup>
+              <optgroup label="Serif">
+                <option value="Playfair Display">Playfair Display</option>
+                <option value="Merriweather">Merriweather</option>
+                <option value="Lora">Lora</option>
+                <option value="PT Serif">PT Serif</option>
+                <option value="Crimson Text">Crimson Text</option>
+              </optgroup>
+              <optgroup label="Monospace">
+                <option value="Roboto Mono">Roboto Mono</option>
+                <option value="Source Code Pro">Source Code Pro</option>
+                <option value="Fira Code">Fira Code</option>
+                <option value="JetBrains Mono">JetBrains Mono</option>
+                <option value="IBM Plex Mono">IBM Plex Mono</option>
+              </optgroup>
+              <optgroup label="Display">
+                <option value="Bebas Neue">Bebas Neue</option>
+                <option value="Oswald">Oswald</option>
+                <option value="Righteous">Righteous</option>
+                <option value="Bangers">Bangers</option>
+                <option value="Permanent Marker">Permanent Marker</option>
+              </optgroup>
+              <optgroup label="Handwriting">
+                <option value="Pacifico">Pacifico</option>
+                <option value="Lobster">Lobster</option>
+                <option value="Dancing Script">Dancing Script</option>
+                <option value="Caveat">Caveat</option>
+                <option value="Satisfy">Satisfy</option>
+              </optgroup>
+            </select>
+            <p style="font-size: 11px; color: var(--text-muted); margin-top: 4px;">
+              Browse more at <a href="https://fonts.google.com" target="_blank" rel="noopener" style="color: var(--accent-blue);">fonts.google.com</a>
+            </p>
+          </div>
+        {/if}
         <div class="field">
           <label for="font-size">Font Size: {fontSize}px</label>
           <input
@@ -648,6 +761,17 @@
         </div>
         <button onclick={handleCheckUpdates}>Check Now</button>
       </section>
+
+      <!-- Transcription Engine -->
+      <section class="settings-section">
+        <h3>Transcription Engine</h3>
+        <p style="font-size: 12px; color: var(--text-secondary); margin-bottom: 12px;">
+          Switch between local (Whisper) and cloud (Deepgram) transcription engines.
+          This will stop the current engine, remove the downloaded files, and restart
+          with the new engine selection.
+        </p>
+        <button class="danger-btn" onclick={handleChangeSidecar}>Change Transcription Engine</button>
+      </section>
     </div>
 
     <div class="settings-footer">
@@ -818,4 +942,18 @@
   .save-message.error {
     color: #f44336;
   }
+
+  .danger-btn {
+    background: transparent;
+    border: 1px solid var(--accent-red, #f44336);
+    color: var(--accent-red, #f44336);
+    padding: 8px 16px;
+    border-radius: 6px;
+    cursor: pointer;
+    font-size: 13px;
+  }
+
+  .danger-btn:hover {
+    background: rgba(244, 67, 54, 0.1);
+  }
 </style>

src/lib/components/SidecarSetup.svelte

@@ -36,11 +36,12 @@
 
     try {
       // Listen for progress events from the Tauri backend
-      unlisten = await listen<{ progress: number; message: string }>(
+      unlisten = await listen<{ downloaded: number; total: number; phase: string; message: string }>(
         "sidecar-download-progress",
         (event) => {
-          progress = event.payload.progress;
+          const { downloaded, total, message } = event.payload;
-          progressMessage = event.payload.message;
+          progress = total > 0 ? (downloaded / total) * 100 : 0;
+          progressMessage = message;
         }
       );
 
@@ -84,11 +85,29 @@
 
     {#if setupState === "choose"}
       <p class="setup-description">
-        The app needs to download its transcription engine before you can start.
+        Choose a transcription engine. You can change this later in Settings.
-        Choose the version that best fits your hardware.
       </p>
 
       <div class="variant-options">
+        <label class="variant-option" class:selected={variant === "cloud"}>
+          <input
+            type="radio"
+            name="variant"
+            value="cloud"
+            bind:group={variant}
+          />
+          <div class="variant-info">
+            <span class="variant-name">Cloud (Deepgram)</span>
+            <span class="variant-desc">~50 MB download</span>
+            <span class="variant-detail">
+              Fast, accurate streaming transcription via Deepgram's servers.
+              Requires internet and a Deepgram API key.
+              Best for most users — low resource usage, works on any hardware.
+            </span>
+            <span class="variant-tag recommended">Recommended</span>
+          </div>
+        </label>
+
         <label class="variant-option" class:selected={variant === "cpu"}>
           <input
             type="radio"
@@ -97,23 +116,16 @@
             bind:group={variant}
           />
           <div class="variant-info">
-            <span class="variant-name">Standard (CPU)</span>
+            <span class="variant-name">Local - CPU</span>
-            <span class="variant-desc">Works on all computers (~500 MB download)</span>
+            <span class="variant-desc">~500 MB download</span>
+            <span class="variant-detail">
+              Runs Whisper AI models locally on your CPU. No internet needed
+              after download. Good for privacy or offline use, but slower and
+              uses more system resources than cloud.
+            </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-name">GPU Accelerated (CUDA)</span>
-            <span class="variant-desc">Faster transcription with NVIDIA GPU (~2 GB download)</span>
-          </div>
-        </label>
       </div>
 
       <button class="download-btn" onclick={startDownload}>
@@ -260,6 +272,30 @@
     color: #888;
   }
 
+  .variant-detail {
+    font-size: 11px;
+    color: #666;
+    line-height: 1.4;
+    margin-top: 2px;
+  }
+
+  .variant-tag {
+    display: inline-block;
+    font-size: 10px;
+    font-weight: 600;
+    text-transform: uppercase;
+    letter-spacing: 0.5px;
+    padding: 2px 6px;
+    border-radius: 3px;
+    margin-top: 4px;
+    width: fit-content;
+  }
+
+  .variant-tag.recommended {
+    background: rgba(76, 175, 80, 0.15);
+    color: #4CAF50;
+  }
+
   .download-btn {
     display: block;
     width: 100%;

src/lib/stores/backend.svelte.ts

@@ -302,6 +302,7 @@ export const backendStore = {
   setPort,
   connect: connectWebSocket,
   disconnect,
+  pollStatus,
   apiUrl,
   apiFetch,
   apiGet,

version.py

@@ -1,7 +1,7 @@
 """Version information for Local Transcription."""
 
-__version__ = "1.4.18"
+__version__ = "2.0.8"
-__version_info__ = (1, 4, 18)
+__version_info__ = (2, 0, 8)
 
 # Version history:
 # 1.4.0 - Auto-update feature: