Connection Status
WebSocket:
Not connected
Session ID:
-
Backend:
-
Latency:
-
Health Check
WebSocket Tests
Used as context/prompt for STT. Leave empty to disable.
Transcription Settings
Improves recognition of domain-specific terms
Advanced Settings
Post-Processing
Uses a local HuggingFace model (fast, free)
Configure LLM in .env: LLM_BASE_URL, LLM_API_KEY, LLM_MODEL
Only correct segments below this confidence
Removes background noise before transcription
Audio Streaming Test
Chunks Sent
0
Bytes Sent
0
Duration
0s
Audio buffer
Buffering...
Transcription starts after ~20s of audio (FFmpeg buffering + 3s Whisper window)
Text-to-Speech Test
Hotword API Test
Transcription Output
Transcriptions will appear here...
Event Log
File Transcription Test
Processing...
0%
Saved Recordings
No recordings saved