Speech-to-Text
Health Check
Check STT service status, loaded models, and available features.
GET
/api/stt/healthReturns service health, loaded models, GPU status, and available features.
No parameters for this endpoint.
Response
{
"status": "ok",
"models": {
"vega": {
"status": "loaded",
"description": "CTC 1B model. 22 Indian + 1100 languages. Optimized for live voice agents.",
"latency": "~100ms"
},
"dhara": {
"status": "loaded",
"description": "Whisper large-v3-turbo. Word timestamps + multilingual.",
"latency": "~400ms"
}
},
"provider": "thinnestai",
"gpu": {
"available": true,
"name": "Tesla T4",
"vram_used_gb": 5.8,
"vram_total_gb": 14.56
},
"features": {
"timestamps": true,
"diarization": true,
"sentiment": true,
"kenlm": true,
"vad_streaming": true,
"keywords_boosting": true,
"pii_redaction": true,
"bgm_filter": true,
"enhance": true,
"denoise": true,
"telephony_detection": true,
"async_transcription": true,
"interim_results": true
}
}Status Values
| Status | Meaning |
|---|---|
ok | All models loaded, service healthy. |
degraded | Some models failed to load. Check models for details. |
error | Service unavailable. |