Generations now return immediately with a 'generating' status and appear
in history right away. TTS runs in a serial background queue to avoid
GPU contention. Users can kick off multiple generations without blocking.
- Async POST /generate creates DB record immediately, queues TTS work
- Serial generation queue prevents concurrent GPU access (Metal/CUDA/CPU)
- SSE endpoint GET /generate/{id}/status for real-time completion tracking
- Retry endpoint POST /generate/{id}/retry for failed generations
- Store engine and model_size on generation records for retry support
- History cards show animated loader (react-loaders) for generating/playing
- Failed generations show retry button instead of actions menu
- Model downloads happen inline in the queue instead of rejecting with 202
- Stale 'generating' records marked as failed on server startup
- Autoplay on generate setting (default: on)
- Show engine name on generation cards
- Remove sidebar generation spinner
- Checkbox alignment fix in settings
- Add model folder migration with byte-level progress tracking (backend + UI)
- Custom models directory support via VOICEBOX_MODELS_DIR env var passed to sidecar
- Hardcoded model descriptions displayed in model detail cards
- Open model folder button in storage location row
- Remove 'not downloaded' badge from model cards
- Fix server settings scroll offset for audio player
- Fix shell open permission to allow file paths
- Add normalize toggle to generation settings
- Split chunking/crossfade sliders into dedicated GenerationSettings card
- Merge connection status badges into ConnectionForm (remove ServerStatus card)
- 2-column grid layout for the entire settings page
- GPU Acceleration: remove icon, badge, and MLX info card
- Models: merge 'Other Voice Models' into single 'Voice Generation' list
- Model detail: remove 'Downloaded' badge, border above actions, swap
badges above stats row, match disk size font to stats
Persisted setting (default 800 chars) controls how long text is split
before generation. Lower values improve quality for long outputs by
keeping each chunk well within the model's context window.
- Slider in Server Connection settings (100–2000 chars, step 50)
- Stored in localStorage via Zustand persist
- Passed as max_chunk_chars on every generation request
- Frontend text limit raised to 50,000 to match backend
Text exceeding max_chunk_chars (default 800) is automatically split at
sentence boundaries, generated per-chunk, and concatenated with a 50ms
crossfade. Works with all engines (Qwen, LuxTTS, Chatterbox, Turbo).
- Abbreviation-aware sentence splitter (Dr., Mr., e.g., decimals)
- CJK sentence-ending punctuation support
- Paralinguistic tag preservation ([laugh], [cough], etc.)
- Per-chunk seed variation to avoid correlated RNG artefacts
- Per-chunk Chatterbox trim (catches hallucination at each boundary)
- max_chunk_chars exposed as per-request param on GenerationRequest
- Text max_length raised to 50,000 characters
Closes#99
- Initialize lastSerializedRef to empty string so first-mount hydration
always runs (fixes initial value not rendering)
- Guard arrow-key menu nav against empty filteredTags (avoids NaN index)
- Disable ARIA role/multiline and detach event handlers when disabled
- Add onBlur to close autocomplete dropdown when editor loses focus
- Chain exception with 'from e' in unload endpoint for better tracebacks
Apply local-var-then-assign pattern to chatterbox_backend.py (multilingual)
to match the turbo backend. Also use _current_model_size fallback in
unload, delete, and status endpoints for consistent Qwen model size checks.
Type / in the text input when using Chatterbox Turbo to open an
autocomplete dropdown with 9 supported paralinguistic tags ([laugh],
[chuckle], [gasp], [cough], [sigh], [groan], [sniff], [shush],
[clear throat]).
- contentEditable div replaces textarea for Turbo engine only
- Tags render as inline styled badges
- Pasting text with [tag] patterns auto-converts to badges
- Badges serialize back to plain [tag] text for the API
- Dropdown portalled to body, opens above caret to avoid overflow
The actual dtype mismatch was in S3Tokenizer.log_mel_spectrogram, not
VoiceEncoder.forward. librosa.load returns float64 numpy, which
torch.from_numpy preserves as double. The STFT output (double) then
hits _mel_filters (float32) in a matmul at s3tokenizer.py:163.
Now patching both entry points after model load:
1. S3Tokenizer.log_mel_spectrogram — cast audio to float32 before STFT
2. VoiceEncoder.forward — cast mels to float32 before LSTM
Remove debug traceback logging (no longer needed).
- POST /models/{model_name}/unload — unloads a specific model from
memory without deleting from disk, supports all engine types
- Frontend: Unload button in model detail dialog when model is loaded
- Delete button remains disabled while loaded (unload first)
The previous approach of patching librosa.load didn't work because
melspectrogram itself performs float64 math (numpy dot, signal.lfilter)
regardless of input dtype. The actual mismatch happens when pack()
creates a float64 tensor from the mel arrays and passes it into the
float32 LSTM weights in VoiceEncoder.forward().
Fix by monkey-patching VoiceEncoder.forward() to call mels.float()
before the LSTM, ensuring the input always matches the model dtype.
soundfile cannot infer format from .tmp extension, causing all
generations to fail with 'No format specified and unable to get
format from file extension'
The upstream VoiceEncoder's melspectrogram only casts to float32 when
hp.normalized_mels is True (it defaults to False), so librosa's float64
output flows through as double tensors into float32 model weights,
causing 'expected m1 and m2 to have the same dtype, but got: float !=
double'. Fix by monkey-patching prepare_conditionals in both Chatterbox
and Chatterbox Turbo backends to ensure librosa.load returns float32.
Both PyTorch and MLX backends silently dropped the language parameter —
it was accepted by generate() but never forwarded to the underlying
Qwen3-TTS model, causing it to default to auto-detection which
frequently confuses similar languages (e.g. Portuguese for Spanish).
- Add LANGUAGE_CODE_TO_NAME mapping (ISO 639-1 to full name) to both backends
- PyTorch: pass language= to generate_voice_clone()
- MLX: pass lang_code= to all 4 model.generate() call sites
- Frontend: auto-sync generation form language with selected voice profile
Closes#97
- save_audio() now writes to .tmp then os.replace() for atomic writes
- /generate endpoint catches OSError with specific messages for ENOENT, EACCES, ENOSPC, and BrokenPipeError
- New /health/filesystem endpoint checks directory existence, write permissions, and disk space
- New DirectoryCheck and FilesystemHealthResponse models
Cherry-picked and expanded from #178 (@Vaibhavee89)
Cherry-picked and adapted from PR #89 and #214:
- Linux audio capture via PulseAudio/PipeWire monitor sources (cpal)
- AMD ROCm GPU support: HSA_OVERRIDE_GFX_VERSION env var, ROCm detection
- Whisper Turbo model (openai/whisper-large-v3-turbo) in all endpoints
- Cleaner Whisper language handling via generate_kwargs
- tauri::async_runtime::spawn fix to prevent panic on app shutdown
- Enable Linux (ubuntu-22.04) in release CI matrix
- New ChatterboxTurboTTSBackend wrapping ChatterboxTurboTTS (ResembleAI/chatterbox-turbo)
- English-only 350M model with paralinguistic tag support ([laugh], [cough], [chuckle])
- Bypasses upstream token=True bug by calling snapshot_download(token=None) + from_local()
- Same CPU-on-macOS forcing and torch.load monkey-patching as multilingual backend
- Full engine integration: generate, stream, model status/download/delete endpoints
- Language dropdown now shows only languages supported by the selected engine
- Per-engine language maps: Qwen (10), LuxTTS (en), Chatterbox (23), Turbo (en)
- Auto-switches to English when selecting English-only engines
- Backend language regex expanded to accept all 23 Chatterbox languages
- Add HFProgressTracker to LuxTTS and Chatterbox backends so tqdm-based
file-level download progress reaches the frontend (previously only Qwen
had this, LuxTTS/Chatterbox showed a static spinner)
- Add progress/current/total/filename fields to ActiveDownloadTask so the
/tasks/active polling endpoint carries progress data
- Show inline progress bar + bytes in the model list and detail modal,
poll at 1s during active downloads (5s otherwise)
- Fix GpuAcceleration crash: cudaStatusLoading was referenced before
initialization in its own useQuery declaration
chatterbox-tts 0.1.6 pins numpy<1.26 and torch==2.6 which are
incompatible with Python 3.12+. Install with --no-deps and list
its sub-dependencies explicitly in requirements.txt.
Also removes HFProgressTracker from chatterbox backend to avoid
'generator didn't stop after throw()' errors from tqdm patching.
- New ChatterboxTTSBackend wrapping ChatterboxMultilingualTTS (ResembleAI/chatterbox)
- Supports 23 languages including Hebrew, forces CPU on macOS (MPS issue)
- Monkey-patches torch.load for CPU loading, forces eager attention for compatibility
- trim_tts_output utility cuts trailing silence/hallucination from Chatterbox output
- Full engine integration: /generate, /generate/stream, model status/download/delete
- Hebrew (he) added to supported languages in frontend and backend validation
- Single flat model dropdown extended with Chatterbox option in both generation UIs
- ModelManagement UI groups LuxTTS and Chatterbox under 'Other Voice Models' section
- Add threading lock to get_tts_backend_for_engine() to prevent race
condition where concurrent requests could create duplicate backend
instances (double-checked locking pattern)
- Fix LuxTTS generate: call .detach().cpu() before .numpy() so it
works on GPU/MPS devices, not just CPU
- Store background download tasks in a module-level set to prevent
garbage collection before completion (asyncio.create_task fire-and-
forget pattern)
- Deduplicate cache_key computation in LuxTTS create_voice_prompt
- Prefix unused sr variable with underscore
- Fix silent Zod validation failure when LuxTTS selected (modelSize was
set to 'default' which failed enum validation, preventing form submit)
- Preserve engine, model size, and language after successful generation
instead of resetting to defaults
- Combine engine + model size into one flat dropdown (Qwen3-TTS 1.7B,
Qwen3-TTS 0.6B, LuxTTS) in both FloatingGenerateBox and GenerationForm
- Add linacodec git dep to requirements.txt (uv-only source, pip can't
resolve it from Zipvoice's pyproject.toml)
- Remove redundant transitive deps from requirements.txt
- Quiet the sidecar setup script (was printing misleading instructions)
piper-phonemize has no PyPI wheels — needs custom find-links URL
from k2-fsa.github.io. Removed redundant transitive deps that
Zipvoice already declares.
Adds 'just' as the recommended dev tool: 'just setup' for one-time
install, 'just dev' to run backend + frontend in one terminal.
Updates CONTRIBUTING.md to document just as the primary setup method.
Introduce LuxTTS (ZipVoice) alongside Qwen TTS, enabling users to choose
between engines at generation time. LuxTTS offers fast, English-focused
voice cloning at 48kHz with ~1GB VRAM.
Backend:
- Add LuxTTSBackend with encode_prompt/generate_speech integration
- Multi-engine registry (get_tts_backend_for_engine) replacing singleton
- Engine-prefixed voice prompt cache keys to avoid collisions
- Engine field on GenerationRequest (default 'qwen' for backward compat)
- Engine dispatch in /generate and /generate/stream endpoints
- LuxTTS in model status, download, and delete maps
Frontend:
- TTS Engine selector dropdown in GenerationForm (Qwen TTS / LuxTTS)
- Conditionally hide Model Size and Delivery Instructions for LuxTTS
- Engine field added to TypeScript types and Zod schema
- LuxTTS section in Model Management page
- Use YAML block scalar for inline run with colons (build-cuda.yml)
- Explicitly set VOICEBOX_BACKEND_VARIANT=cpu instead of setdefault (server.py)
- Use Path.replace() for atomic move on all platforms (cuda_download.py)
- Log actual exception in checksum fetch warning (cuda_download.py)
Add the ability to download a CUDA-enabled backend binary (~2.4 GB) and
swap it in via a backend-only restart, solving the #1 user pain point
(19 open 'GPU not detected' issues caused by GitHub's 2 GB asset limit).
Backend:
- cuda_download.py: download from R2 (primary) or GitHub split-parts
(fallback), SHA-256 verification, atomic writes, progress via SSE
- 4 new endpoints: GET/POST/DELETE /backend/cuda-*, GET cuda-progress
- server.py: --version flag, auto-detect variant from binary name
- build_binary.py: --cuda flag for CUDA PyInstaller builds
- split_binary.py: split large binaries into <2GB GitHub Release assets
- CI workflow for building CUDA binary
Tauri:
- restart_server command (stop -> wait -> start)
- start_server prefers CUDA binary from {data_dir}/backends/ if present
- Version mismatch check: runs --version before launching CUDA binary
Frontend:
- GpuAcceleration component: download, progress, restart, switch, delete
- API client + types for CUDA status and management
- Platform lifecycle: restartServer() on Tauri/Web
- Aggressive 1s health polling during restart for fast reconnection
- Updated caching methods in MLX, PyTorch, and backend to ensure models are fully downloaded before being marked as cached.
- Improved progress tracking to filter out non-download progress and provide accurate feedback during model downloads.
- Enhanced HFProgressTracker to skip non-byte progress bars and ensure meaningful progress reporting.
- Refactored progress initialization to provide immediate feedback while fetching metadata from HuggingFace.
- Added error handling and logging for better debugging during cache checks and download processes.
- Rearranged imports for consistency in useModelDownloadToast hook.
- Improved logging in useModelDownloadToast for better debugging during download events.
- Updated progress calculation to handle cases where progress exceeds 100%.
- Enhanced toast notifications to reflect download completion and error states.
- Introduced throttling in ProgressManager to optimize SSE updates and prevent overwhelming clients.
- Added new test scripts for monitoring SSE events during model downloads, ensuring accurate progress reporting.
- Rearranged imports for consistency across components.
- Enhanced the ModelManagement component to include detailed logging for download actions and errors.
- Updated the ModelProgress component to connect to SSE only when actively downloading, preventing connection exhaustion.
- Added a downloading state to the model status to indicate ongoing downloads.
- Improved toast notifications for model downloads with completion and error callbacks.
- Refactored the useModelDownloadToast hook to support new callbacks for download completion and error handling.
- Updated backend model status to reflect downloading state during active downloads.
- Added a step to install PyTorch with CUDA for Windows in the release workflow.
- Updated model references in backend/main.py to use openai/whisper models instead of mlx-community for the MLX backend.
- Bumped version numbers for @voicebox/app, @voicebox/landing, @voicebox/tauri, and @voicebox/web to 0.1.11.
- Added a new `useAutoUpdater` hook to check for app updates on startup and notify users with toast messages.
- Enhanced `UpdateStatus` component to handle version retrieval errors more gracefully.
- Updated dependencies in `package.json` for Tauri plugins to support new update functionalities.
- Added a default `type` prop set to 'button' in the CircleButton component to ensure proper button behavior.
- Enhanced the component's flexibility by allowing the type to be overridden through props.
- Implemented delete confirmation dialogs for both HistoryTable and SampleList components to enhance user experience and prevent accidental deletions.
- Added state management for handling the selected item to be deleted and the visibility of the delete dialog.
- Refactored delete handling functions to utilize the new dialog confirmation flow, improving code clarity and maintainability.
- Added `clear_profile_cache` function to manage cache files for specific profiles.
- Integrated cache clearing in `add_profile_sample`, `delete_profile`, and `delete_profile_sample` functions to ensure stale audio caches are invalidated after modifications.
- Enhanced `clear_voice_prompt_cache` to also delete combined audio files, improving overall cache management.
- Introduced a new directory for manual test scripts aimed at debugging and validating backend functionality.
- Added README.md detailing the purpose and usage of various test scripts, including tests for TTS generation, model downloads, and progress tracking.
- Included an __init__.py file to define the test suite structure and provide context for the tests.
- Introduced methods to check if models are cached locally in MLX and PyTorch backends.
- Enhanced progress tracking during model loading to filter out non-download progress when models are cached.
- Updated HFProgressTracker to conditionally report progress based on download status.
- Added test scripts for monitoring SSE events during model downloads and verifying progress tracking functionality.
- Improved overall error handling and logging for better debugging during model download processes.
- Updated HistoryTable to implement infinite scrolling for loading history items dynamically.
- Introduced state management for accumulated history and total item count.
- Added Intersection Observer for triggering additional data fetches when scrolling.
- Implemented cache clearing functionality in the backend to manage voice prompt caches effectively.
- Improved loading indicators and user feedback for data fetching states.
- Refactored code for better readability and maintainability.
- Replaced direct Tauri API calls with a unified platform context across multiple components, enhancing code maintainability and readability.
- Removed the deprecated tauri.ts file, consolidating platform-related logic into the new PlatformContext.
- Updated components such as App, AudioPlayer, and ServerSettings to utilize the new platform context for lifecycle management and server interactions.
- Improved platform detection and handling for audio playback and system audio capture functionalities.
- Ensured consistent error handling and user feedback across the application when interacting with platform-specific features.
- Modified the file extension filters in useHistory.ts and useProfiles.ts to only allow 'zip' files, removing 'voicebox.zip' for a more streamlined export process.
- Added user-selected read-write permission in Entitlements.plist to enhance file handling capabilities.
- Updated README.md to highlight MLX backend performance improvements on Mac with Metal acceleration.
- Refined ProfileCard and ProfileForm components by optimizing imports and improving error handling for avatar uploads.
- Adjusted landing page content to better describe features, including a new multi-voice narrative editor and performance optimizations for different platforms.
- Bumped version to 0.1.11 in Cargo.lock to reflect recent changes.
- Updated hidden imports in build_binary.py to replace 'mlx_audio.asr' with 'mlx_audio.stt'.
- Enhanced model loading logic in MLX and PyTorch backends to ensure proper progress tracking during model downloads.
- Improved error handling and context management for progress tracking in both backends.
- Bumped version to 0.1.10 in Cargo.lock to reflect recent changes.
- Added support for MLX backend on Apple Silicon, enabling optimized performance for TTS and STT tasks.
- Implemented platform detection to dynamically select between MLX and PyTorch based on the runtime environment.
- Updated build process to include MLX-specific dependencies and configurations for macOS.
- Refactored backend code to improve model loading and inference logic, accommodating backend-specific requirements.
- Enhanced documentation to clarify backend selection and performance benefits for different platforms.
- Streamlined installation instructions and troubleshooting guidance for MLX-related issues.
- Added support for MLX backend on Apple Silicon, enabling optimized performance for TTS and STT tasks.
- Updated release workflow to include MLX-specific dependencies and configurations for macOS platforms.
- Refactored backend code to dynamically select between MLX and PyTorch based on the runtime environment.
- Enhanced model loading and inference logic to accommodate backend-specific requirements, including updated model IDs and hidden imports.
- Improved health check and model status reporting to reflect the active backend type.
- Streamlined caching mechanisms to support both backend types, ensuring compatibility and performance.
- Updated MLX_AUDIO.md to reflect validated status and included detailed validation results, model mapping, and API usage examples.
- Added a demo script (demo.py) for testing audio generation speed and functionality.
- Introduced a test script (test_tts.py) to validate MLX audio model loading and generation, ensuring robust testing for future developments.
- Created a .gitignore file in the mlx-test directory to exclude unnecessary files from version control.
- Introduced a new backend for MLX audio to enable GPU acceleration on macOS Apple Silicon, improving performance and user experience.
- Implemented platform detection to switch between MLX and PyTorch backends based on the runtime environment.
- Added new streaming capabilities for TTS and STT, enhancing real-time audio generation.
- Updated API endpoints and frontend components to support new features while maintaining backward compatibility.
- Created documentation for backend integration and performance comparisons.
- Removed unnecessary isDownloading variable and related logic.
- Consolidated model download state reset to the finally block for improved clarity and reliability.
- Enhanced error handling by ensuring model download state is reset in case of failure.
- Reorganized import statements for clarity and consistency.
- Adjusted formatting of state declarations for better readability.
- Streamlined JSX structure for improved visual hierarchy.
- Updated dialog descriptions for consistency in presentation.
- Made minor adjustments to spacing and layout for enhanced UI consistency.
- Changed border radius of the "No stories yet" message to rounded-2xl for a softer appearance.
- Updated story item borders to rounded-2xl to enhance visual cohesion across the component.
- Updated button styles in ProfileForm for better visual consistency and user experience.
- Replaced Pencil icon with Edit in SampleList for clearer action representation.
- Introduced CircleButton component for action buttons in SampleList, enhancing UI responsiveness and clarity.
- Improved layout and hover effects for action buttons in SampleList to streamline user interactions.
- Added functionality to upload, delete, and retrieve avatar images for voice profiles.
- Introduced new API endpoints for avatar management, including upload and delete operations.
- Enhanced profile forms and components to support avatar image handling, including previews and error handling.
- Updated database schema to include avatar_path for profiles and added necessary migrations.
- Implemented image validation and processing utilities to ensure proper avatar uploads.
- Reorganized import statements for clarity.
- Enhanced conditional checks for restoring saved files with improved formatting.
- Streamlined draft saving logic by consolidating variable declarations.
- Updated UI components for better structure and readability in the form layout.
- Created a new .npmrc file to enforce bun usage.
- Bumped version numbers for multiple packages to 0.1.9 in bun.lock.
- Added react-sound-visualizer dependency to enhance audio visualization features.
- Introduced convert:assets script in package.json for asset optimization.
- Updated CONTRIBUTING.md with instructions for converting assets to web formats.
- Added documentation files for API endpoints and developer guidelines in the docs directory.
- Added thread-safe mechanisms to the ProgressManager for handling model download progress updates.
- Introduced a main event loop setter to ensure safe operations from background threads.
- Improved listener notification to handle updates in a thread-safe manner.
- Updated methods to ensure thread safety when accessing progress data.
- Updated API endpoints from `/api/...` to `/...` for consistency.
- Modified the speech generation example to include a language parameter.
- Revised the profile creation example to use JSON format instead of form data.
- Introduced a flag to track download status in useGenerationForm, ensuring proper UI updates during model downloads.
- Updated ProgressManager to only send initial progress updates if the model is actively downloading or extracting, preventing outdated status messages from being sent.
- Improved error handling and logging for better visibility into model download processes.
- Introduced functionality to save and restore form state as a draft when creating a new voice profile.
- Added helper functions for converting files to and from base64 format to facilitate file handling.
- Updated the API types to use a more flexible LanguageCode type for language parameters.
- Enhanced the UI store to manage profile form drafts, improving user experience during profile creation.
- Added a useEffect hook to automatically scroll the timeline when the playhead moves past the halfway point of the visible area, enhancing user experience during playback.
- Introduced a constant for TIME_RULER_HEIGHT to improve code readability.
- Updated drag position calculations to subtract the time ruler height, ensuring accurate positioning of clips relative to the tracks area.
- Updated CONTRIBUTING.md to include instructions for building with a local Qwen3-TTS development version, facilitating easier testing and development.
- Refactored FloatingGenerateBox component to streamline the rendering of text and instruct fields, improving code readability and maintainability.
- Added functionality to handle auto-resizing of text areas based on content changes, enhancing user experience.
- Improved event handling for keyboard interactions in StoryTrackEditor, allowing for play/pause functionality with the spacebar.
- Introduced a MiniSamplePlayer component in SampleList for better audio playback control, including play, pause, and seek features.
- Implemented sample update functionality in the backend, allowing users to edit reference text for audio samples, with appropriate error handling and user feedback.
- Updated the health check endpoint to include the type of GPU available (CUDA or MPS).
- Modified the HealthResponse model to accommodate the new gpu_type field, enhancing the response with detailed GPU information.
- This change improves the clarity of system capabilities for users and developers.
- Added a new `/shutdown` endpoint to allow graceful server shutdown via HTTP.
- Implemented process tree management functions to handle child processes during shutdown on Windows.
- Updated the `stop_server` function to attempt graceful shutdown before forcefully terminating processes.
- Enhanced error handling and logging for shutdown operations.
- Implemented background tasks for downloading TTS and Whisper models to prevent blocking HTTP responses.
- Enhanced error handling during model downloads, providing users with real-time feedback on download status.
- Updated HTTP responses to indicate when models are being downloaded, improving user experience during model initialization.
- Moved model download logic into a separate asynchronous function to allow non-blocking HTTP responses.
- Improved error handling by tracking download status and reporting errors without interrupting the main request flow.
- The frontend is now expected to poll the progress endpoint for download status updates.
Jamie Pine