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
- HistoryTable: skip row key handler when focus is on Actions button (Enter/Space)
- StoryList: expose selected story (aria-pressed, 'Selected' in label)
- ProfileCard: skip card key handler when focus is on Export/Edit/Delete
- VoicesTab: keep table semantics; edit button in first cell instead of role=button on row
- PR-ACCESSIBILITY.md: 'Fine-tune' wording, 'focus on the text area' phrasing
Made-with: Cursor
- Audio player: aria-labels for Play/Pause, Loop, Mute, Close; labelled playback and volume sliders
- Generation: aria-labels for Generate speech and Fine tune instructions buttons
- Voice cards: focusable, labelled, Enter/Space to select
- History rows: focusable, labelled, Enter/Space to play; transcript textarea labelled
- Voices tab: focusable rows, labelled, Enter/Space to edit; Actions button labelled
- Model management: focusable model rows and labelled Download/Delete buttons
- Server tab: regions with aria-label and tabIndex for Connection, Status, App Updates
- Stories: focusable story rows, labelled, Enter/Space to select; Actions and track editor buttons labelled
- Voice profile samples: Play/Pause/Stop and mini-player slider labelled
Tested with NVDA and Narrator on Windows. See docs/PR-ACCESSIBILITY.md for full description.
Made-with: Cursor
- Fix transcribe_audio to use whisper-large-v3 mapping (not openai/whisper-large)
- Propagate error field in progress-only fallback path for get_active_tasks
- Use removed return value in cancel endpoint to vary response message
- Add error rollback to handleCancel with toast on failure
- Make isCancelling per-model instead of global
- Fix inverted chevron icons in Problems panel
- Move all clears under lock in clear_all_tasks
- Simplify cancel_download to use dict.pop()
- Add cancel (X) button on downloading and errored model items
- Add collapsible Problems panel (VS Code-style) showing error details
- Add "Clear All" button to reset all stale download/error state
- Add POST /models/download/cancel endpoint to dismiss individual downloads
- Add POST /tasks/clear endpoint to reset all task and progress state
- Include error messages in /tasks/active response for visibility
- Capture SSE error messages client-side for immediate display
- Fix whisper-large using wrong HF repo (openai/whisper-large → openai/whisper-large-v3)
- Fix Whisper HF repo mapping in both PyTorch and MLX backends
- Shorten error toast to point users to Problems panel instead of wall of text
Address CodeRabbit review feedback:
- Makefile: only set WEBKIT_DISABLE_DMABUF_RENDERER=1 when running on
Linux with an NVIDIA GPU detected via lspci
- main.rs: validate webview origin before auto-granting microphone
permission — only allow for trusted local origins (tauri://, localhost,
127.0.0.1)
Co-Authored-By: Claude Opus 4.6 <[email protected]>
WebKitGTK denies getUserMedia by default. This adds webkit2gtk as a
Linux dependency and configures the webview to enable media streams
and auto-grant UserMediaPermissionRequest for microphone access.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
WebKitGTK fails to create GBM buffers with NVIDIA proprietary drivers,
resulting in an empty/blank Tauri window. Set WEBKIT_DISABLE_DMABUF_RENDERER=1
in the dev target to work around this.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
- Add validation in create_profile() to check for existing names before insert
- Add validation in update_profile() to prevent renaming to duplicate names
- Improve error handling in API endpoints with user-friendly messages
- Add comprehensive test suite for duplicate name validation
- Update CHANGELOG.md with fix details
This fix prevents database constraint violations and provides clear
error messages when users attempt to create or update profiles with
names that already exist in the database.
Implements offline mode patch for API stability issues:
- Add hf_offline_patch.py to monkey-patch huggingface_hub
- Force cache-only lookups before mlx_audio imports
- Create symlink from original Qwen repo to MLX community version
when only MLX version is cached
This fixes:
- Issue #150: Internet required even with cached models
- Issue #151: API crashes when HF network fails
The patch ensures that if models are locally cached, no network
requests are made to HuggingFace during speech generation.
Adds a persistent top-of-page banner linking to spacebot.sh,
another project by the creator of Voicebox. Uses existing design
tokens for a consistent look.
Exposes the existing remote server mode through a checkbox in Server
Connection settings. When enabled, the server binds to 0.0.0.0 instead
of 127.0.0.1, making it accessible from other devices on the network.
The plumbing already existed (Rust sidecar passes --host 0.0.0.0 when
remote=true, serverStore has mode state, Python backend accepts --host),
but the UI hardcoded startServer(false). This wires it up.
Closes#104
The web version was missing the Tailwind CSS Vite plugin, causing
CSS to not load at all. This adds the same plugin configuration
that exists in the tauri version.
Fixes#121
The /generate endpoint created the voice prompt before loading the
user's requested model size. Since create_voice_prompt() internally
calls load_model_async(None), it fell back to the hardcoded default
of "1.7B", causing the 1.7B model to be downloaded even when the
user explicitly selected 0.6B.
This reorders the operations so the requested model is loaded first,
ensuring create_voice_prompt() and generate() use the correct model.
Co-Authored-By: Claude Opus 4.6 <[email protected]>
The distributed macOS aarch64 binary shipped without MLX acceleration despite
the model and backend code supporting it. Two root causes:
1. **OSError not caught in platform_detect.py**
PyInstaller bundles isolate the filesystem, so when MLX tries to load its
Metal shader libraries (.metallib) it raises OSError, not ImportError.
platform_detect.get_backend_type() only caught ImportError, causing a
silent fallback to PyTorch even on Apple Silicon hardware.
Fix: broaden the except clause to (ImportError, OSError, RuntimeError)
and import mlx.core instead of mlx (forces native lib loading eagerly).
2. **collect_data_files used instead of collect_all for MLX**
build_binary.py and voicebox-server.spec used --collect-data /
collect_data_files for mlx and mlx_audio. This copies Python source and
pure-Python data, but NOT native shared libraries (.dylib, .metallib).
Fix: switch to --collect-all / collect_all which captures binaries too,
then pass them to Analysis(binaries=...) in the spec.
Result: macOS Apple Silicon users now get MLX inference (~4-5x faster than
PyTorch CPU), matching the performance documented in the README.
20 tests covering:
- All 6 default local origins are allowed
- Arbitrary external origins are blocked
- Preflight (OPTIONS) requests respect the allowlist
- VOICEBOX_CORS_ORIGINS env var extends the allowlist
- Edge cases: empty env, whitespace trimming, trailing commas
Tests use a minimal FastAPI app mirroring the real CORS config,
so they run without ML dependencies (torch, numpy, etc.).
The wildcard `allow_origins=["*"]` allows any website the user visits to
make requests to the local voicebox backend, potentially triggering TTS
generation or reading voice profiles without consent.
Restrict to the known Tauri webview and Vite dev server origins by
default. Users running in remote server mode can set
VOICEBOX_CORS_ORIGINS to allow additional origins.
The export endpoints (export-audio, export generation, export profile,
export story) crash with `'latin-1' codec can't encode characters` when
the generated text or profile/story name contains non-ASCII characters
(e.g. Cyrillic, Chinese, Arabic).
Root cause: Python's `str.isalnum()` passes Unicode letters through to
the filename, but HTTP headers are encoded as latin-1 by the ASGI server,
which cannot represent characters outside the 0-255 range.
Fix: introduce `_safe_content_disposition()` helper that builds a
standards-compliant header with an ASCII-only `filename` fallback and a
RFC 5987 `filename*=UTF-8''...` parameter for Unicode-capable clients.
Fixes#68
Co-authored-by: Cursor <[email protected]>
James Pine