Files
openchamber/packages/web/server/lib/dictation/local/model-catalog.js
T
Bohdan Triapitsyn de1b85ac56 feat(voice): first-class voice input and local TTS across web, desktop, and mobile (#2018)
Complete rebuild of voice input on a server-authoritative streaming
architecture, replacing the legacy Web Speech / whole-blob / WASM engines
and the dead voice-agent layer (~4k lines removed).

Speech-to-text (dictation):
- Client streams 16 kHz mono PCM16 chunks over /api/dictation/ws with
  seq/ack ordering; buffered audio is retained and replayed on reconnect
- Server transcribes and streams live partial transcripts back;
  segments auto-commit every ~15s with silence suppression and adaptive
  finalization timeouts
- Local provider (default, zero config): sherpa-onnx models in a forked
  worker process — auto-download with progress, staged extraction with
  verification, corrupt-model auto-recovery, idle shutdown after 5 min
- Model catalog with settings picker (accuracy/speed ratings, sizes,
  download/delete): Parakeet TDT v2 (English) and v3 (25 European
  languages, auto-detected), Whisper base and tiny (multilingual, light)
- OpenAI-compatible provider for any Whisper endpoint
- Composer overlay with live transcript, volume meter, timer, and
  cancel / insert / insert-and-send actions; failed transcriptions keep
  their audio for retry or accepting the partial text as-is
- Configurable keyboard shortcut (default mod+alt+v) toggles dictation;
  Enter confirms and Escape cancels while recording
- Overlay is pixel-aligned with the composer (measured footer height,
  matching paddings/typography/gaps) — no layout shift when toggling

Text-to-speech:
- Local Kokoro provider (English, 11 voices) synthesized in the same
  worker via /api/dictation/tts/speak, managed by the shared model
  pipeline; sentence-pipelined playback keeps time-to-first-audio at
  ~1 sentence regardless of message length, and stop cancels in-flight
  synthesis
- Sanitizer keeps inline-code content (strips backticks only), reads
  interword slashes aloud, and removes only absolute file paths

Settings:
- Voice page unified: a single read-aloud toggle owns all playback
  options (the confusing "Enable Voice Mode" is gone); a new "Enable
  voice input" toggle (default on, persisted to settings.json) hides
  the composer mic entirely when disabled

Mobile and transport:
- iOS/Android microphone permissions added (dictation was previously
  impossible on mobile)
- Fixed Android WebSocket upgrades: the Capacitor WebView origin
  (https://localhost) was missing from the packaged-client allowlist,
  403-ing every WS connection — root cause of the old mobile SSE lock,
  which is now removed for all transports

Security and conventions:
- All HTTP routes sit behind the global /api auth gate; the WS upgrade
  explicitly validates the UI session and origin, with oc_url_token
  narrowly allowlisted and covered by tests; the dictation socket mints
  a fresh URL token before connecting
- Routes register before the generic OpenCode proxy; the client goes
  through runtimeFetch/getRuntimeUrlResolver, and runtime switches
  reset the dictation socket
- VS Code deliberately reports dictation as unavailable (no server
  process in that runtime)

CI: workflow Node bumped 20 -> 22 to match the repo engines and fix
better-sqlite3 installs broken by node-gyp@latest on Node 20.

New dependency: sherpa-onnx-node (prebuilt N-API; macOS/Linux x64+arm64,
Windows x64 — Windows-on-ARM falls back to the OpenAI-compatible provider)
2026-07-04 02:48:07 +03:00

142 lines
4.4 KiB
JavaScript

/**
* Catalog of local sherpa-onnx STT models available for dictation.
* Models are downloaded on demand from the k2-fsa GitHub releases and
* extracted under the OpenChamber speech-models directory.
*
* `type` selects the recognizer construction path in the worker:
* - 'nemo_transducer': encoder/decoder/joiner transducer (Parakeet)
* - 'whisper': encoder/decoder Whisper export
* `files` maps logical roles to file names inside the extracted directory.
*/
import path from 'path';
export const LOCAL_STT_MODEL_CATALOG = {
'parakeet-tdt-0.6b-v2-int8': {
type: 'nemo_transducer',
archiveUrl:
'https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8.tar.bz2',
extractedDir: 'sherpa-onnx-nemo-parakeet-tdt-0.6b-v2-int8',
files: {
encoder: 'encoder.int8.onnx',
decoder: 'decoder.int8.onnx',
joiner: 'joiner.int8.onnx',
tokens: 'tokens.txt',
},
description: 'NVIDIA Parakeet TDT v2 (English)',
},
'parakeet-tdt-0.6b-v3-int8': {
type: 'nemo_transducer',
archiveUrl:
'https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8.tar.bz2',
extractedDir: 'sherpa-onnx-nemo-parakeet-tdt-0.6b-v3-int8',
files: {
encoder: 'encoder.int8.onnx',
decoder: 'decoder.int8.onnx',
joiner: 'joiner.int8.onnx',
tokens: 'tokens.txt',
},
description: 'NVIDIA Parakeet TDT v3 (25 European languages, auto-detected)',
},
'whisper-base-int8': {
type: 'whisper',
archiveUrl:
'https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-base.tar.bz2',
extractedDir: 'sherpa-onnx-whisper-base',
files: {
encoder: 'base-encoder.int8.onnx',
decoder: 'base-decoder.int8.onnx',
tokens: 'base-tokens.txt',
},
description: 'OpenAI Whisper base (multilingual, smaller and lighter)',
},
'whisper-tiny-int8': {
type: 'whisper',
archiveUrl:
'https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.tar.bz2',
extractedDir: 'sherpa-onnx-whisper-tiny',
files: {
encoder: 'tiny-encoder.int8.onnx',
decoder: 'tiny-decoder.int8.onnx',
tokens: 'tiny-tokens.txt',
},
description: 'OpenAI Whisper tiny (multilingual, fastest and lightest)',
},
};
/**
* Local text-to-speech models (sherpa-onnx OfflineTts). Downloaded and
* managed through the same pipeline as the STT models.
*/
export const LOCAL_TTS_MODEL_CATALOG = {
'kokoro-en-v0_19': {
type: 'kokoro',
archiveUrl:
'https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/kokoro-en-v0_19.tar.bz2',
extractedDir: 'kokoro-en-v0_19',
files: {
model: 'model.onnx',
voices: 'voices.bin',
tokens: 'tokens.txt',
espeakData: 'espeak-ng-data',
},
description: 'Kokoro TTS (English, natural voices)',
},
};
export const DEFAULT_LOCAL_STT_MODEL = 'parakeet-tdt-0.6b-v2-int8';
export const DEFAULT_LOCAL_TTS_MODEL = 'kokoro-en-v0_19';
export const LOCAL_STT_MODEL_IDS = Object.keys(LOCAL_STT_MODEL_CATALOG);
export const LOCAL_TTS_MODEL_IDS = Object.keys(LOCAL_TTS_MODEL_CATALOG);
/**
* @param {string} modelId
* @returns {boolean}
*/
export function isLocalSttModelId(modelId) {
return typeof modelId === 'string' && Object.hasOwn(LOCAL_STT_MODEL_CATALOG, modelId);
}
/**
* @param {string} modelId
* @returns {boolean}
*/
export function isLocalTtsModelId(modelId) {
return typeof modelId === 'string' && Object.hasOwn(LOCAL_TTS_MODEL_CATALOG, modelId);
}
/**
* Any managed local model (STT or TTS).
* @param {string} modelId
* @returns {boolean}
*/
export function isLocalModelId(modelId) {
return isLocalSttModelId(modelId) || isLocalTtsModelId(modelId);
}
/**
* Spec lookup across both catalogs (STT and TTS).
* @param {string} modelId
*/
export function getLocalSttModelSpec(modelId) {
const spec = LOCAL_STT_MODEL_CATALOG[modelId] ?? LOCAL_TTS_MODEL_CATALOG[modelId];
if (!spec) {
throw new Error(`Unknown local speech model id: ${modelId}`);
}
return {
id: modelId,
...spec,
requiredFiles: Object.values(spec.files),
};
}
/**
* @param {string} modelsDir
* @param {string} modelId
* @returns {string}
*/
export function getLocalSttModelDir(modelsDir, modelId) {
return path.join(modelsDir, getLocalSttModelSpec(modelId).extractedDir);
}