Configure STT language from settings (#941)
* Configure STT language from settings * Update websocket speech mock for language config
This commit is contained in:
@@ -161,6 +161,7 @@ In the mobile app, enter the password in the direct connection setup screen.
|
||||
- `PASEO_LOCAL_MODELS_DIR`, control local model directory
|
||||
- `PASEO_DICTATION_LOCAL_STT_MODEL`, override local dictation STT model
|
||||
- `PASEO_VOICE_LOCAL_STT_MODEL`, `PASEO_VOICE_LOCAL_TTS_MODEL`, override local voice STT/TTS models
|
||||
- `PASEO_DICTATION_LANGUAGE`, `PASEO_VOICE_LANGUAGE`, override dictation and voice STT language
|
||||
- `PASEO_VOICE_LOCAL_TTS_SPEAKER_ID`, `PASEO_VOICE_LOCAL_TTS_SPEED`, optional local voice TTS tuning
|
||||
|
||||
## Schema
|
||||
|
||||
Reference in New Issue
Block a user