chore: bump sidecar version to 1.0.5 [skip ci]

chore: bump version to 2.0.6 [skip ci]
Fix YAML parse error: use block scalar for echo with colons
2026-04-08 00:43:01 +00:00 · 2026-04-08 00:37:28 +00:00 · 2026-04-07 17:21:42 -07:00 · 2026-04-08 00:19:25 +00:00 · 2026-04-07 17:16:36 -07:00 · 2026-04-07 17:14:23 -07:00
27 changed files with 758 additions and 171 deletions
--- a/.gitea/workflows/build-app-linux.yml
+++ b/.gitea/workflows/build-app-linux.yml
@@ -13,10 +13,11 @@ jobs:
    runs-on: ubuntu-latest
    env:
      NODE_VERSION: "20"
-      RELEASE_TAG: ${{ inputs.tag }}
+      RELEASE_TAG: "${{ inputs.tag }}"
    steps:
      - name: Show tag
-        run: echo "Building for tag: ${RELEASE_TAG}"
+        run: |
+          echo "Building for tag: ${RELEASE_TAG}"

      - uses: actions/checkout@v4
        with:
--- a/.gitea/workflows/build-app-macos.yml
+++ b/.gitea/workflows/build-app-macos.yml
@@ -13,10 +13,11 @@ jobs:
    runs-on: macos-latest
    env:
      NODE_VERSION: "20"
-      RELEASE_TAG: ${{ inputs.tag }}
+      RELEASE_TAG: "${{ inputs.tag }}"
    steps:
      - name: Show tag
-        run: echo "Building for tag: ${RELEASE_TAG}"
+        run: |
+          echo "Building for tag: ${RELEASE_TAG}"

      - uses: actions/checkout@v4
        with:
--- a/.gitea/workflows/build-app-windows.yml
+++ b/.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
--- a/.gitea/workflows/build-sidecar-cloud.yml
+++ b/.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
--- a/.gitea/workflows/build-sidecar-linux.yml
+++ b/.gitea/workflows/build-sidecar-linux.yml
@@ -13,10 +13,11 @@ jobs:
    runs-on: ubuntu-latest
    env:
      PYTHON_VERSION: "3.11"
-      RELEASE_TAG: ${{ inputs.tag }}
+      RELEASE_TAG: "${{ inputs.tag }}"
    steps:
      - name: Show tag
-        run: echo "Building for tag: ${RELEASE_TAG}"
+        run: |
+          echo "Building for tag: ${RELEASE_TAG}"

      - uses: actions/checkout@v4
        with:
--- a/.gitea/workflows/build-sidecar-macos.yml
+++ b/.gitea/workflows/build-sidecar-macos.yml
@@ -13,10 +13,11 @@ jobs:
    runs-on: macos-latest
    env:
      PYTHON_VERSION: "3.11"
-      RELEASE_TAG: ${{ inputs.tag }}
+      RELEASE_TAG: "${{ inputs.tag }}"
    steps:
      - name: Show tag
-        run: echo "Building for tag: ${RELEASE_TAG}"
+        run: |
+          echo "Building for tag: ${RELEASE_TAG}"

      - uses: actions/checkout@v4
        with:
--- a/.gitea/workflows/build-sidecar-windows.yml
+++ b/.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
--- a/.gitea/workflows/release.yml
+++ b/.gitea/workflows/release.yml
@@ -1,14 +1,7 @@
 name: Release

 on:
-  push:
-    branches: [main]
-    paths:
-      - 'src/**'
-      - 'src-tauri/**'
-      - 'package.json'
-      - 'vite.config.ts'
-      - 'index.html'
+  workflow_dispatch:

 jobs:
  test:
@@ -116,50 +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

-      - name: Clean up old app releases
-        env:
-          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
-        run: |
-          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
-          KEEP=3
-          PROTECT_TAG="v1.4.0"
-
-          echo "Cleaning up old app releases (keeping latest ${KEEP} + ${PROTECT_TAG})..."
-
-          # Get all app releases (v* tags, not sidecar-v*)
-          RELEASES=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
-            "${REPO_API}/releases?limit=50" | jq -c '[.[] | select(.tag_name | startswith("v")) | select(.tag_name | startswith("sidecar") | not)]')
-
-          TOTAL=$(echo "$RELEASES" | jq 'length')
-          echo "Found ${TOTAL} app releases"
-
-          if [ "$TOTAL" -le "$KEEP" ]; then
-            echo "Nothing to clean up"
-            exit 0
-          fi
-
-          # Skip the newest KEEP releases, delete the rest (except protected)
-          echo "$RELEASES" | jq -c ".[$KEEP:][]" | while read -r release; do
-            ID=$(echo "$release" | jq -r '.id')
-            TAG=$(echo "$release" | jq -r '.tag_name')
-
-            if [ "$TAG" = "$PROTECT_TAG" ]; then
-              echo "  Protecting ${TAG}"
-              continue
-            fi
-
-            echo "  Deleting release ${TAG} (ID: ${ID})..."
-            curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" \
-              "${REPO_API}/releases/${ID}"
-            # Keep the git tag -- only delete the release (assets).
-            # Deleting tags breaks builds that haven't checked out yet.
-          done
-          echo "Cleanup complete"
+      # NOTE: Automatic cleanup disabled -- it races with async builds.
+      # Clean up old releases manually from the Gitea UI when needed.
--- a/.gitea/workflows/sidecar-release.yml
+++ b/.gitea/workflows/sidecar-release.yml
@@ -1,14 +1,6 @@
 name: Sidecar Release

 on:
-  push:
-    branches: [main]
-    paths:
-      - 'client/**'
-      - 'server/**'
-      - 'backend/**'
-      - 'pyproject.toml'
-      - 'local-transcription-headless.spec'
  workflow_dispatch:

 jobs:
@@ -126,7 +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}" \
@@ -136,37 +128,5 @@ jobs:
            echo "  -> HTTP ${HTTP_CODE}"
          done

-      - name: Clean up old sidecar releases
-        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}"
-          KEEP=2
-
-          echo "Cleaning up old sidecar releases (keeping latest ${KEEP})..."
-
-          # Get all sidecar releases (sidecar-v* tags)
-          RELEASES=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
-            "${REPO_API}/releases?limit=50" | jq -c '[.[] | select(.tag_name | startswith("sidecar-v"))]')
-
-          TOTAL=$(echo "$RELEASES" | jq 'length')
-          echo "Found ${TOTAL} sidecar releases"
-
-          if [ "$TOTAL" -le "$KEEP" ]; then
-            echo "Nothing to clean up"
-            exit 0
-          fi
-
-          # Skip the newest KEEP releases, delete the rest
-          echo "$RELEASES" | jq -c ".[$KEEP:][]" | while read -r release; do
-            ID=$(echo "$release" | jq -r '.id')
-            TAG=$(echo "$release" | jq -r '.tag_name')
-
-            echo "  Deleting sidecar release ${TAG} (ID: ${ID})..."
-            curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" \
-              "${REPO_API}/releases/${ID}"
-            # Keep the git tag -- only delete the release (assets).
-            # Deleting tags breaks builds that haven't checked out yet.
-          done
-          echo "Cleanup complete"
+      # NOTE: Automatic cleanup disabled -- it races with async builds.
+      # Clean up old releases manually from the Gitea UI when needed.
--- a/backend/app_controller.py
+++ b/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.transcription_engine_realtime import RealtimeTranscriptionEngine, TranscriptionResult
+from client.models import 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 +94,18 @@ 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

        # State
        self._state = AppState.INITIALIZING
@@ -243,15 +259,12 @@ 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 +297,18 @@ 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:
+                from client.transcription_engine_realtime import RealtimeTranscriptionEngine as _RTE
+                RealtimeTranscriptionEngine = _RTE
+
+            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,
@@ -577,12 +602,18 @@ 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 +627,7 @@ 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 +671,13 @@ 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:
-            devices.append({"id": dev_id, "name": dev_name})
+        if self.device_manager:
+            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 ────────────────────────────────────────────
--- a/backend/tests/test_api_server.py
+++ b/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 = [
--- a/backend/tests/test_app_controller.py
+++ b/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

--- a/client/deepgram_transcription.py
+++ b/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)
-            except Empty:
+                # Use run_in_executor to avoid blocking the async event loop
+                # (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:
--- a/client/models.py
+++ b/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}"
--- a/client/transcription_engine_realtime.py
+++ b/client/transcription_engine_realtime.py
@@ -8,30 +8,8 @@ from threading import Lock
 import logging


-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}"
+# Re-export TranscriptionResult from the shared models module for backward compatibility
+from client.models import TranscriptionResult  # noqa: F401

    def to_dict(self) -> dict:
        """Convert to dictionary."""
--- a/local-transcription-cloud.spec
+++ b/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',
+)
--- a/package.json
+++ b/package.json
@@ -1,7 +1,7 @@
 {
  "name": "local-transcription",
  "private": true,
-  "version": "2.0.3",
+  "version": "2.0.6",
  "type": "module",
  "scripts": {
    "dev": "vite dev",
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "local-transcription"
-version = "1.0.4"
+version = "1.0.5"
 description = "A standalone desktop application for real-time speech-to-text transcription using Whisper models"
 readme = "README.md"
 requires-python = ">=3.9"
--- a/src-tauri/Cargo.lock
+++ b/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",
--- a/src-tauri/Cargo.toml
+++ b/src-tauri/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "local-transcription"
-version = "2.0.3"
+version = "2.0.6"
 description = "Real-time speech-to-text transcription for streamers"
 authors = ["Local Transcription Contributors"]
 edition = "2021"
--- a/src-tauri/gen/schemas/capabilities.json
+++ b/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"]}}
--- a/src-tauri/src/lib.rs
+++ b/src-tauri/src/lib.rs
@@ -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!())
--- a/src-tauri/src/sidecar/mod.rs
+++ b/src-tauri/src/sidecar/mod.rs
@@ -554,18 +554,27 @@ impl SidecarManager {
    // -- private helpers -------------------------------------------------------

    fn build_dev_command(&self) -> Result<std::process::Command, String> {
-        let mut cmd = std::process::Command::new("python");
-        cmd.args(["-u", "-m", "backend.main_headless"]); // -u = unbuffered
+        // Use `uv run` to ensure we use the project's venv, not system Python
+        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)
-        if let Some(dirs) = DIRS.get() {
-            let project_root = dirs
-                .resource_dir
-                .parent() // src-tauri
-                .and_then(|p| p.parent()); // project root
-            if let Some(root) = project_root {
-                cmd.current_dir(root);
-            }
+        // Find the project root: try CARGO_MANIFEST_DIR first (set at compile time),
+        // then fall back to resource_dir parent chain
+        let manifest_dir = option_env!("CARGO_MANIFEST_DIR").map(std::path::PathBuf::from);
+        let project_root = manifest_dir
+            .as_ref()
+            .and_then(|d| d.parent()) // src-tauri -> project root
+            .or_else(|| {
+                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");
@@ -676,6 +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
 // ---------------------------------------------------------------------------
--- a/src-tauri/tauri.conf.json
+++ b/src-tauri/tauri.conf.json
@@ -1,6 +1,6 @@
 {
  "productName": "Local Transcription",
-  "version": "2.0.3",
+  "version": "2.0.6",
  "identifier": "net.anhonesthost.local-transcription",
  "build": {
    "frontendDist": "../dist",
--- a/src/lib/components/Settings.svelte
+++ b/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>
--- a/src/lib/components/SidecarSetup.svelte
+++ b/src/lib/components/SidecarSetup.svelte
@@ -84,11 +84,29 @@

    {#if setupState === "choose"}
      <p class="setup-description">
-        The app needs to download its transcription engine before you can start.
-        Choose the version that best fits your hardware.
+        Choose a transcription engine. You can change this later in Settings.
      </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"
@@ -97,8 +115,13 @@
            bind:group={variant}
          />
          <div class="variant-info">
-            <span class="variant-name">Standard (CPU)</span>
-            <span class="variant-desc">Works on all computers (~500 MB download)</span>
+            <span class="variant-name">Local - CPU</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>

@@ -110,8 +133,13 @@
            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>
+            <span class="variant-name">Local - GPU (NVIDIA CUDA)</span>
+            <span class="variant-desc">~2 GB download</span>
+            <span class="variant-detail">
+              Runs Whisper AI models locally using your NVIDIA GPU for fast
+              transcription. No internet needed after download. Requires an
+              NVIDIA GPU with CUDA support.
+            </span>
          </div>
        </label>
      </div>
@@ -260,6 +288,30 @@
    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%;
--- a/version.py
+++ b/version.py
@@ -1,7 +1,7 @@
 """Version information for Local Transcription."""

-__version__ = "2.0.3"
-__version_info__ = (2, 0, 3)
+__version__ = "2.0.6"
+__version_info__ = (2, 0, 6)

 # Version history:
 # 1.4.0 - Auto-update feature:
Author	SHA1	Message	Date
Gitea Actions	375669f657	chore: bump sidecar version to 1.0.5 [skip ci]	2026-04-08 00:43:01 +00:00
Gitea Actions	c8b11fb0ad	chore: bump version to 2.0.6 [skip ci]	2026-04-08 00:37:28 +00:00
Developer	273a926f03	Fix YAML parse error: use block scalar for echo with colons All checks were successful Tests / Python Backend Tests (push) Successful in 5s Details Tests / Frontend Tests (push) Successful in 7s Details Tests / Rust Sidecar Tests (push) Successful in 2m7s Details Gitea's YAML parser treats `echo "text: value"` as a mapping when on a single `run:` line. Using block scalar (`run: \|`) avoids this. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 17:21:42 -07:00
Gitea Actions	5bbbc38875	chore: bump version to 2.0.5 [skip ci]	2026-04-08 00:19:25 +00:00
Developer	d50be6654d	Fix dispatch failures and disable automatic cleanup All checks were successful Tests / Python Backend Tests (push) Successful in 5s Details Tests / Frontend Tests (push) Successful in 8s Details Tests / Rust Sidecar Tests (push) Successful in 2m8s Details 1. Quote RELEASE_TAG env vars in all workflow files. Unquoted ${{ inputs.tag }} caused YAML parse errors on some Gitea runners, making dispatch return HTTP 500 for Linux/macOS. 2. Disable automatic release cleanup in both coordinators. The cleanup races with async builds -- it deletes the release before builds finish uploading their assets. Clean up old releases manually from the Gitea UI instead. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 17:16:36 -07:00
Developer	68abf49018	Log dispatch error responses for debugging Some checks failed Tests / Python Backend Tests (push) Successful in 5s Details Tests / Frontend Tests (push) Successful in 7s Details Tests / Rust Sidecar Tests (push) Has been cancelled Details Show the Gitea API response body when dispatch returns non-204, to help diagnose why Linux/macOS dispatches return HTTP 500. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 17:14:23 -07:00
Gitea Actions	8cc2a3ec7a	chore: bump version to 2.0.4 [skip ci]	2026-04-08 00:09:39 +00:00
Developer	8aa9dfc644	Update Cargo.lock and generated Tauri schemas All checks were successful Tests / Python Backend Tests (push) Successful in 5s Details Tests / Frontend Tests (push) Successful in 7s Details Tests / Rust Sidecar Tests (push) Successful in 2m10s Details Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 17:03:40 -07:00
Developer	3f16aa838d	Add ability to change transcription engine from Settings Some checks failed Tests / Python Backend Tests (push) Successful in 6s Details Tests / Frontend Tests (push) Successful in 8s Details Tests / Rust Sidecar Tests (push) Has been cancelled Details New features: - Settings > Transcription Engine > "Change Transcription Engine" button stops the sidecar, deletes downloaded files, and reloads the app to show the engine selection screen - Improved SidecarSetup descriptions with detailed explanations of each variant and "Recommended" tag on Cloud (Deepgram) - Cloud option listed first as the recommended choice - New reset_sidecar Tauri command that cleans up sidecar files Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 17:02:31 -07:00
Developer	3d3d7ec3c5	Add cloud-only sidecar variant (~50MB vs 500MB-2GB) All checks were successful Tests / Python Backend Tests (push) Successful in 6s Details Tests / Frontend Tests (push) Successful in 7s Details Tests / Rust Sidecar Tests (push) Successful in 1m59s Details Lightweight Deepgram-only sidecar that excludes PyTorch, faster-whisper, RealtimeSTT, and CUDA. Only includes audio capture + WebSocket streaming to Deepgram. Requires a Deepgram API key (BYOK or managed mode). Changes: - client/models.py: Extracted TranscriptionResult into standalone module so deepgram_transcription.py doesn't transitively import torch - backend/app_controller.py: Made RealtimeTranscriptionEngine and DeviceManager imports lazy (only loaded when remote.mode == "local") - local-transcription-cloud.spec: PyInstaller spec excluding all ML deps - SidecarSetup.svelte: Added "Cloud Only (Deepgram)" variant option - build-sidecar-cloud.yml: CI workflow building cloud sidecar for all 3 OS - sidecar-release.yml: Dispatches cloud build alongside CPU/CUDA builds Sidecar download options are now: - Standard (CPU): ~500 MB - local Whisper on any computer - GPU Accelerated (CUDA): ~2 GB - local Whisper with NVIDIA GPU - Cloud Only (Deepgram): ~50 MB - requires API key, no local models Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 16:57:43 -07:00
Developer	bb039399fc	Add font source/family settings matching v1.4.0 feature set All checks were successful Tests / Python Backend Tests (push) Successful in 6s Details Tests / Frontend Tests (push) Successful in 8s Details Tests / Rust Sidecar Tests (push) Successful in 2m11s Details Restored the font configuration that was missing from the Tauri rewrite. Settings now include: - Font Source: System Font, Web-Safe, Google Font - System Font: text input for any installed font family - Web-Safe: dropdown with 13 universal fonts (Arial, Courier New, etc.) - Google Font: dropdown with 35 fonts organized by category (Sans Serif, Serif, Monospace, Display, Handwriting) - Font Size: range slider (8-32px) All font settings are saved to config and applied to the OBS web display and server sync. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 16:40:52 -07:00
Developer	9dcb14e92c	Fix Deepgram streaming latency All checks were successful Tests / Python Backend Tests (push) Successful in 5s Details Tests / Frontend Tests (push) Successful in 9s Details Tests / Rust Sidecar Tests (push) Successful in 2m5s Details Three changes to reduce transcription delay: 1. Send loop: queue.get() was blocking the asyncio event loop, stalling the receive loop and delaying transcription results. Now uses run_in_executor() to avoid blocking the event loop. 2. Block size: reduced from 4096 (~256ms) to 1024 (~64ms) for more frequent, smaller audio chunks. Deepgram handles streaming better with smaller packets. 3. Added punctuate=true and smart_format=true to Deepgram BYOK params for cleaner transcription output. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 16:31:50 -07:00
Developer	8db9b8298b	Fix dev mode sidecar launch and engine reload on mode change All checks were successful Tests / Python Backend Tests (push) Successful in 6s Details Tests / Frontend Tests (push) Successful in 7s Details Tests / Rust Sidecar Tests (push) Successful in 1m57s Details 1. Dev mode: use `uv run python` instead of bare `python` to ensure the project venv is used. Also use CARGO_MANIFEST_DIR to find the project root reliably. 2. Engine reload: changing remote.mode (local/managed/byok) now triggers a full engine reload. Previously only model and device changes triggered reload, so switching to Deepgram had no effect until the app was restarted. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 16:25:07 -07:00
Developer	411779f578	Make release and sidecar-release manual-only while testing All checks were successful Tests / Python Backend Tests (push) Successful in 4s Details Tests / Frontend Tests (push) Successful in 7s Details Tests / Rust Sidecar Tests (push) Successful in 1m59s Details Removed push triggers from both coordinator workflows. They now only run via workflow_dispatch (manual "Run workflow" button). Re-enable push triggers once the build pipeline is stable. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 16:04:06 -07:00
Developer	bc6055a707	Add workflow_dispatch trigger to release.yml Some checks failed Tests / Python Backend Tests (push) Successful in 6s Details Tests / Frontend Tests (push) Successful in 8s Details Tests / Rust Sidecar Tests (push) Has been cancelled Details Allows manually triggering app releases from the Gitea Actions UI. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-07 16:03:18 -07:00