chore: bump sidecar version to 1.0.8 [skip ci]

Audio capture started immediately after spawning the WebSocket thread,
but the WebSocket hadn't connected yet. Audio chunks sent to the
unconnected WebSocket caused a broken pipe error.

Fix: added a threading.Event that start_recording() waits on (up to
15s) before opening the audio stream. The event is set in _ws_lifecycle
after the WebSocket connects and handshake completes.

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

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

@@ -40,26 +40,17 @@ 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)
         run: |
-          rm -rf dist/local-transcription-backend build/
+          uv sync --no-sources
+          # PyPI's default torch on Linux includes CUDA (~800MB).
+          # Replace with CPU-only torch from the dedicated index.
           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:

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

@@ -54,29 +54,19 @@ 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
         run: |
-          Remove-Item -Recurse -Force dist\local-transcription-backend, build -ErrorAction SilentlyContinue
+          $env:UV_NO_SOURCES = "1"
+          uv sync
+          # PyPI's default torch includes CUDA. Replace with CPU-only.
           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/cleanup-releases.yml

@@ -0,0 +1,102 @@
+name: Cleanup Old Releases
+
+on:
+  workflow_dispatch:
+    inputs:
+      keep_app_releases:
+        description: 'Number of app releases to keep'
+        required: false
+        default: '3'
+      keep_sidecar_releases:
+        description: 'Number of sidecar releases to keep'
+        required: false
+        default: '2'
+      dry_run:
+        description: 'Dry run (show what would be deleted without deleting)'
+        required: false
+        default: 'true'
+
+jobs:
+  cleanup:
+    name: Cleanup Old Releases
+    runs-on: ubuntu-latest
+    steps:
+      - name: Cleanup releases
+        env:
+          BUILD_TOKEN: ${{ secrets.BUILD_TOKEN }}
+        run: |
+          REPO_API="${GITHUB_SERVER_URL}/api/v1/repos/${GITHUB_REPOSITORY}"
+          KEEP_APP="${{ inputs.keep_app_releases }}"
+          KEEP_SIDECAR="${{ inputs.keep_sidecar_releases }}"
+          DRY_RUN="${{ inputs.dry_run }}"
+
+          echo "=== Cleanup Configuration ==="
+          echo "Keep app releases: ${KEEP_APP}"
+          echo "Keep sidecar releases: ${KEEP_SIDECAR}"
+          echo "Dry run: ${DRY_RUN}"
+          echo ""
+
+          # Fetch all releases
+          ALL_RELEASES=$(curl -s -H "Authorization: token ${BUILD_TOKEN}" \
+            "${REPO_API}/releases?limit=50")
+
+          # ── App releases (v* tags, not sidecar-v*) ──
+          echo "=== App Releases ==="
+          APP_RELEASES=$(echo "$ALL_RELEASES" | jq -c '[.[] | select(.tag_name | startswith("v")) | select(.tag_name | startswith("sidecar") | not)]')
+          APP_TOTAL=$(echo "$APP_RELEASES" | jq 'length')
+          echo "Found ${APP_TOTAL} app releases, keeping ${KEEP_APP}"
+
+          if [ "$APP_TOTAL" -gt "$KEEP_APP" ]; then
+            echo "$APP_RELEASES" | jq -c ".[$KEEP_APP:][]" | while read -r release; do
+              ID=$(echo "$release" | jq -r '.id')
+              TAG=$(echo "$release" | jq -r '.tag_name')
+              SIZE=$(echo "$release" | jq '[.assets[]?.size // 0] | add // 0')
+              SIZE_MB=$(echo "scale=1; $SIZE / 1048576" | bc 2>/dev/null || echo "?")
+
+              # Protect v1.4.0 (last pre-Tauri release)
+              if [ "$TAG" = "v1.4.0" ]; then
+                echo "  PROTECT ${TAG} (${SIZE_MB} MB)"
+                continue
+              fi
+
+              if [ "$DRY_RUN" = "true" ]; then
+                echo "  WOULD DELETE ${TAG} (ID: ${ID}, ${SIZE_MB} MB)"
+              else
+                echo "  DELETING ${TAG} (ID: ${ID}, ${SIZE_MB} MB)..."
+                curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" \
+                  "${REPO_API}/releases/${ID}"
+              fi
+            done
+          else
+            echo "  Nothing to clean up"
+          fi
+
+          echo ""
+
+          # ── Sidecar releases (sidecar-v* tags) ──
+          echo "=== Sidecar Releases ==="
+          SIDECAR_RELEASES=$(echo "$ALL_RELEASES" | jq -c '[.[] | select(.tag_name | startswith("sidecar-v"))]')
+          SIDECAR_TOTAL=$(echo "$SIDECAR_RELEASES" | jq 'length')
+          echo "Found ${SIDECAR_TOTAL} sidecar releases, keeping ${KEEP_SIDECAR}"
+
+          if [ "$SIDECAR_TOTAL" -gt "$KEEP_SIDECAR" ]; then
+            echo "$SIDECAR_RELEASES" | jq -c ".[$KEEP_SIDECAR:][]" | while read -r release; do
+              ID=$(echo "$release" | jq -r '.id')
+              TAG=$(echo "$release" | jq -r '.tag_name')
+              SIZE=$(echo "$release" | jq '[.assets[]?.size // 0] | add // 0')
+              SIZE_MB=$(echo "scale=1; $SIZE / 1048576" | bc 2>/dev/null || echo "?")
+
+              if [ "$DRY_RUN" = "true" ]; then
+                echo "  WOULD DELETE ${TAG} (ID: ${ID}, ${SIZE_MB} MB)"
+              else
+                echo "  DELETING ${TAG} (ID: ${ID}, ${SIZE_MB} MB)..."
+                curl -s -X DELETE -H "Authorization: token ${BUILD_TOKEN}" \
+                  "${REPO_API}/releases/${ID}"
+              fi
+            done
+          else
+            echo "  Nothing to clean up"
+          fi
+
+          echo ""
+          echo "=== Done ==="

README.md

@@ -267,6 +267,15 @@ Both workflows require a `BUILD_TOKEN` secret in the repo settings (Gitea API to
 
 ## Troubleshooting
 
+### macOS: "App is damaged and can't be opened"
+macOS Gatekeeper blocks unsigned applications. Since the app is not yet signed with an Apple Developer certificate, you need to remove the quarantine flag before opening:
+
+```bash
+xattr -cr "/Applications/Local Transcription.app"
+```
+
+Then open the app normally. You only need to do this once after downloading.
+
 ### Model Loading Issues
 - Models download automatically on first use to `~/.cache/huggingface/`
 - First run requires internet connection

backend/app_controller.py

@@ -106,6 +106,12 @@ class AppController:
                 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
@@ -300,8 +306,17 @@ class AppController:
             # Lazy-import heavy local transcription dependencies
             global RealtimeTranscriptionEngine
             if RealtimeTranscriptionEngine is None:
-                from client.transcription_engine_realtime import RealtimeTranscriptionEngine as _RTE
-                RealtimeTranscriptionEngine = _RTE
+                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)
@@ -358,7 +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 ──────────────────────────────────────
 
@@ -373,7 +396,14 @@ class AppController:
         try:
             success = self.transcription_engine.start_recording()
             if not success:
-                return False, "Failed to start recording"
+                import logging
+                # Check if there's a recent error in the logger
+                err_detail = getattr(self.transcription_engine, '_last_error', '')
+                msg = f"Failed to start recording"
+                if err_detail:
+                    msg += f": {err_detail}"
+                print(f"ERROR: {msg}")
+                return False, msg
 
             # Start server sync if enabled
             if self.config.get('server_sync.enabled', False):

client/deepgram_transcription.py

@@ -156,17 +156,30 @@ class DeepgramTranscriptionEngine:
             return True
 
         self._stop_event.clear()
+        self._ws_connected = threading.Event()
         self._is_recording = True
 
         # Start the asyncio event-loop thread (handles WS send/receive)
         self._thread = threading.Thread(target=self._run_event_loop, daemon=True)
         self._thread.start()
 
+        # Wait for the WebSocket to connect before starting audio capture.
+        # Without this, audio chunks arrive before the WS is open -> broken pipe.
+        if not self._ws_connected.wait(timeout=15):
+            logger.error("Timed out waiting for Deepgram WebSocket connection")
+            print("ERROR: Timed out waiting for Deepgram WebSocket connection")
+            self._last_error = "Timed out connecting to Deepgram"
+            self._is_recording = False
+            self._stop_event.set()
+            return False
+
         # Start the audio capture stream
         try:
             self._start_audio_stream()
         except Exception as exc:
             logger.error("Failed to open audio stream: %s", exc)
+            print(f"ERROR: Failed to open audio stream: {exc}")
+            self._last_error = f"Audio stream error: {exc}"
             self._is_recording = False
             self._stop_event.set()
             return False
@@ -283,6 +296,11 @@ class DeepgramTranscriptionEngine:
                 if not await self._managed_handshake():
                     return
 
+            # Signal that the WebSocket is connected and ready
+            logger.info("WebSocket connected to Deepgram")
+            if hasattr(self, '_ws_connected'):
+                self._ws_connected.set()
+
             # Run send and receive concurrently
             await asyncio.gather(
                 self._send_loop(),

package.json

@@ -1,7 +1,7 @@
 {
   "name": "local-transcription",
   "private": true,
-  "version": "2.0.6",
+  "version": "2.0.9",
   "type": "module",
   "scripts": {
     "dev": "vite dev",

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "local-transcription"
-version = "1.0.5"
+version = "1.0.8"
 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.toml

@@ -1,6 +1,6 @@
 [package]
 name = "local-transcription"
-version = "2.0.6"
+version = "2.0.9"
 description = "Real-time speech-to-text transcription for streamers"
 authors = ["Local Transcription Contributors"]
 edition = "2021"

src-tauri/src/lib.rs

@@ -71,6 +71,17 @@ pub fn run() {
             sidecar::reset_sidecar,
             write_log,
         ])
-        .run(tauri::generate_context!())
-        .expect("error while running tauri application");
+        .build(tauri::generate_context!())
+        .expect("error while building tauri application")
+        .run(|app, event| {
+            if let tauri::RunEvent::Exit = event {
+                // Stop the sidecar when the app exits
+                if let Some(state) = app.try_state::<sidecar::ManagedSidecar>() {
+                    if let Ok(mut mgr) = state.0.lock() {
+                        eprintln!("[app] Stopping sidecar on exit...");
+                        mgr.stop();
+                    }
+                }
+            }
+        });
 }

src-tauri/tauri.conf.json

@@ -1,6 +1,6 @@
 {
   "productName": "Local Transcription",
-  "version": "2.0.6",
+  "version": "2.0.9",
   "identifier": "net.anhonesthost.local-transcription",
   "build": {
     "frontendDist": "../dist",

src/lib/components/Controls.svelte

@@ -8,17 +8,25 @@
   );
   let isLoading = $state(false);
 
+  let errorMessage = $state("");
+
   async function toggleTranscription() {
     if (isLoading) return;
     isLoading = true;
+    errorMessage = "";
     try {
       if (isTranscribing) {
         await backendStore.apiPost("/api/stop");
       } else {
         await backendStore.apiPost("/api/start");
       }
-    } catch (err) {
-      console.error("Failed to toggle transcription:", err);
+      // Poll status to update UI immediately instead of waiting
+      // for WebSocket broadcast (which can be delayed or missed)
+      await backendStore.pollStatus();
+    } catch (err: unknown) {
+      const msg = err instanceof Error ? err.message : String(err);
+      console.error("Failed to toggle transcription:", msg);
+      errorMessage = msg;
     } finally {
       isLoading = false;
     }
@@ -101,9 +109,19 @@
   <button onclick={saveTranscriptions} disabled={!backendStore.connected}>
     Save
   </button>
+
+  {#if errorMessage}
+    <span class="error-msg">{errorMessage}</span>
+  {/if}
 </div>
 
 <style>
+  .error-msg {
+    color: #f44336;
+    font-size: 12px;
+    margin-left: 8px;
+  }
+
   .controls {
     display: flex;
     align-items: center;

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;
-          progressMessage = event.payload.message;
+          const { downloaded, total, message } = event.payload;
+          progress = total > 0 ? (downloaded / total) * 100 : 0;
+          progressMessage = message;
         }
       );
 
@@ -125,23 +126,6 @@
           </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">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>
 
       <button class="download-btn" onclick={startDownload}>

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__ = "2.0.6"
-__version_info__ = (2, 0, 6)
+__version__ = "2.0.9"
+__version_info__ = (2, 0, 9)
 
 # Version history:
 # 1.4.0 - Auto-update feature: