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>
Start Transcription button now shows the error message when it fails
instead of silently reverting. Common causes:
- Missing PortAudio library on Linux
- Audio device not accessible
- Deepgram connection failure
Also added error details to backend console output and captured
the last error from the Deepgram engine for better diagnostics.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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>
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>
Test suite covering all three layers:
Python backend (25 tests):
- AppController: state machine, start/stop, callbacks, settings reload
- API server: REST endpoints, config CRUD, status, devices
- Config: dot-notation get/set, persistence, nested paths
- Main headless: ready event port format validation
Svelte frontend (14 tests via Vitest):
- Backend store: exported properties/methods, port derivation, URLs
- Config store: method names (fetchConfig not loadConfig), defaults
- Transcriptions store: add/clear/plaintext
- File extension regression: ensures $state runes only in .svelte.ts
Rust sidecar (24 tests via cargo test):
- Platform/arch detection, asset name construction
- Ready event deserialization (with extra fields tolerance)
- Path construction, version read/write, old version cleanup
- Zip extraction, SidecarManager lifecycle
CI workflow (.gitea/workflows/test.yml):
- Runs on push to main and PRs
- Three parallel jobs: Python, Frontend, Rust
Also fixes three bugs found during test planning:
- Settings: /api/check-updates -> GET /api/check-update
- Settings: /api/remote/login -> /api/login
- Settings: /api/remote/register -> /api/register
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
New files:
- client/deepgram_transcription.py — DeepgramTranscriptionEngine with
managed mode (proxy) and BYOK mode (direct Deepgram). Sends raw binary
PCM audio over WebSocket, handles both proxy and Deepgram response formats.
Modified files:
- config/default_config.yaml — Replace remote_processing with new remote
section (mode, server_url, auth_token, byok_api_key, deepgram_model, language)
- client/config.py — Add migration from old remote_processing config
- gui/settings_dialog_qt.py — Replace Remote Processing group with
Transcription Mode section (Local/Managed/BYOK radio buttons, login/register
dialogs, balance display, model selector)
- gui/main_window_qt.py — Select engine based on remote.mode config,
add error and credits_low handlers
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add UpdateChecker class to query Gitea API for latest releases
- Show update dialog with release notes when new version available
- Open browser to release page for download (handles large files)
- Allow users to skip specific versions or defer updates
- Add "Check for Updates Now" button in settings
- Check automatically on startup (respects 24-hour interval)
- Pre-configured for repo.anhonesthost.net/streamer-tools
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add color settings (user_color, text_color, background_color) to config
- Add color picker buttons in Settings dialog with alpha support for backgrounds
- Update local web display to use configurable colors
- Send per-user colors with transcriptions to multi-user server
- Update Node.js server to apply per-user colors on display page
- Improve server landing page: replace tech details with display options reference
- Bump version to 1.3.2
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Font changes:
- Consolidate font settings into single Display Settings section
- Support Web-Safe, Google Fonts, and Custom File uploads for both displays
- Fix Google Fonts URL encoding (use + instead of %2B for spaces)
- Fix per-speaker font inline style quote escaping in Node.js display
- Add font debug logging to help diagnose font issues
- Update web server to sync all font settings on settings change
- Remove deprecated PHP server documentation files
New features:
- Add remote transcription service for GPU offloading
- Add instance lock to prevent multiple app instances
- Add version tracking
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Major refactor to eliminate word loss issues using RealtimeSTT with
dual-layer VAD (WebRTC + Silero) instead of time-based chunking.
## Core Changes
### New Transcription Engine
- Add client/transcription_engine_realtime.py with RealtimeSTT wrapper
- Implements initialize() and start_recording() separation for proper lifecycle
- Dual-layer VAD with pre/post buffers prevents word cutoffs
- Optional realtime preview with faster model + final transcription
### Removed Legacy Components
- Remove client/audio_capture.py (RealtimeSTT handles audio)
- Remove client/noise_suppression.py (VAD handles silence detection)
- Remove client/transcription_engine.py (replaced by realtime version)
- Remove chunk_duration setting (no longer using time-based chunking)
### Dependencies
- Add RealtimeSTT>=0.3.0 to pyproject.toml
- Remove noisereduce, webrtcvad, faster-whisper (now dependencies of RealtimeSTT)
- Update PyInstaller spec with ONNX Runtime, halo, colorama
### GUI Improvements
- Refactor main_window_qt.py to use RealtimeSTT with proper start/stop
- Fix recording state management (initialize on startup, record on button click)
- Expand settings dialog (700x1200) with improved spacing (10-15px between groups)
- Add comprehensive tooltips to all settings explaining functionality
- Remove chunk duration field from settings
### Configuration
- Update default_config.yaml with RealtimeSTT parameters:
- Silero VAD sensitivity (0.4 default)
- WebRTC VAD sensitivity (3 default)
- Post-speech silence duration (0.3s)
- Pre-recording buffer (0.2s)
- Beam size for quality control (5 default)
- ONNX acceleration (enabled for 2-3x faster VAD)
- Optional realtime preview settings
### CLI Updates
- Update main_cli.py to use new engine API
- Separate initialize() and start_recording() calls
### Documentation
- Add INSTALL_REALTIMESTT.md with migration guide and benefits
- Update INSTALL.md: Remove FFmpeg requirement (not needed!)
- Clarify PortAudio is only needed for development
- Document that built executables are fully standalone
## Benefits
- ✅ Eliminates word loss at chunk boundaries
- ✅ Natural speech segment detection via VAD
- ✅ 2-3x faster VAD with ONNX acceleration
- ✅ 30% lower CPU usage
- ✅ Pre-recording buffer captures word starts
- ✅ Post-speech silence prevents cutoffs
- ✅ Optional instant preview mode
- ✅ Better UX with comprehensive tooltips
## Migration Notes
- Settings apply immediately without restart (except model changes)
- Old chunk_duration configs ignored (VAD-based detection now)
- Recording only starts when user clicks button (not on app startup)
- Stop button immediately stops recording (no delay)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
**Model Reload Fixes:**
- Properly disconnect signals before reconnecting to prevent duplicate connections
- Wait for previous model loader thread to finish before starting new one
- Add garbage collection after unloading model to free memory
- Improve error handling in model reload callback
**Settings Dialog:**
- Remove duplicate success message (callback handles it)
- Only show message if no callback is defined
**Transcription Engine:**
- Explicitly delete model reference before setting to None
- Force garbage collection to ensure memory is freed
This prevents crashes when switching models, especially when done
multiple times in succession or while the app is under load.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Phase 2 implementation: Multiple streamers can now merge their captions
into a single stream using a PHP server.
PHP Server (server/php/):
- server.php: API endpoint for sending/streaming transcriptions
- display.php: Web page for viewing merged captions in OBS
- config.php: Server configuration
- .htaccess: Security settings
- README.md: Comprehensive deployment guide
Features:
- Room-based isolation (multiple groups on same server)
- Passphrase authentication per room
- Real-time streaming via Server-Sent Events (SSE)
- Different colors for each user
- File-based storage (no database required)
- Auto-cleanup of old rooms
- Works on standard PHP hosting
Client-Side:
- client/server_sync.py: HTTP client for sending to PHP server
- Settings dialog updated with server sync options
- Config updated with server_sync section
Server Configuration:
- URL: Server endpoint (e.g., http://example.com/transcription/server.php)
- Room: Unique room name for your group
- Passphrase: Shared secret for authentication
OBS Integration:
Display URL format:
http://example.com/transcription/display.php?room=ROOM&passphrase=PASS&fade=10×tamps=true
NOTE: Main window integration pending (client sends transcriptions)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Changes:
1. Changed UI text from "Recording" to "Transcribing" for clarity
2. Implemented overlapping audio chunks to prevent word cutoff
Audio Overlap Feature:
- Added overlap_duration parameter (default: 0.5 seconds)
- Audio chunks now overlap by 0.5s to capture words at boundaries
- Prevents missed words when chunks are processed separately
- Configurable via audio.overlap_duration in config.yaml
How it works:
- Each 3-second chunk includes 0.5s from the previous chunk
- Buffer advances by (chunk_size - overlap_size) instead of full chunk
- Ensures words at chunk boundaries are captured in at least one chunk
- No duplicate transcription due to Whisper's context handling
Example with 3s chunks and 0.5s overlap:
Chunk 1: [0.0s - 3.0s]
Chunk 2: [2.5s - 5.5s] <- 0.5s overlap
Chunk 3: [5.0s - 8.0s] <- 0.5s overlap
Files modified:
- client/audio_capture.py: Implemented overlapping buffer logic
- config/default_config.yaml: Added overlap_duration setting
- gui/main_window_qt.py: Updated UI text, passed overlap param
- main_cli.py: Passed overlap param
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Phase 1 Complete - Standalone Desktop Application
Features:
- Real-time speech-to-text with Whisper (faster-whisper)
- PySide6 desktop GUI with settings dialog
- Web server for OBS browser source integration
- Audio capture with automatic sample rate detection and resampling
- Noise suppression with Voice Activity Detection (VAD)
- Configurable display settings (font, timestamps, fade duration)
- Settings apply without restart (with automatic model reloading)
- Auto-fade for web display transcriptions
- CPU/GPU support with automatic device detection
- Standalone executable builds (PyInstaller)
- CUDA build support (works on systems without CUDA hardware)
Components:
- Audio capture with sounddevice
- Noise reduction with noisereduce + webrtcvad
- Transcription with faster-whisper
- GUI with PySide6
- Web server with FastAPI + WebSocket
- Configuration system with YAML
Build System:
- Standard builds (CPU-only): build.sh / build.bat
- CUDA builds (universal): build-cuda.sh / build-cuda.bat
- Comprehensive BUILD.md documentation
- Cross-platform support (Linux, Windows)
Documentation:
- README.md with project overview and quick start
- BUILD.md with detailed build instructions
- NEXT_STEPS.md with future enhancement roadmap
- INSTALL.md with setup instructions
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
