Realtime Transcription API Beta
Real-time speech recognition powered by relax.ai and NVIDIA.
API Docs
Get API Key
Disconnected
Start Recording
Press
Space
to start
Advanced
Turn Detection
Mode
Manual (click Stop)
Silence Detection
Semantic VAD
Threshold
0.5
Prefix Padding
ms
Silence Duration
ms
Eagerness
Low (wait longer)
Medium (balanced)
High (quick response)
Click Stop when you're done speaking.
Streaming Mode
Latency
80ms (Fastest)
160ms (Low latency)
560ms (Balanced)
1120ms (High accuracy)
Advanced Latency
Interim Interval
ms
Chunk Duration
ms
Features
Show performance stats
Noise cancellation
Coming soon
Transcript
Copy
Clear
Press Start Recording to begin