Realtime Transcription API Beta

Real-time speech recognition powered by relax.ai and NVIDIA.

First-Time Setup

Microphone access requires a secure context. Since this site uses HTTP, you need to enable it in your browser settings.

Chrome

  1. Open chrome://flags/#unsafely-treat-insecure-origin-as-secure
  2. Enter this site's URL in the text box
  3. Change dropdown to Enabled
  4. Click Relaunch at the bottom

Firefox

  1. Open about:config
  2. Search for media.getusermedia.insecure.enabled
  3. Set it to true

How to Use

  1. Click Start Recording and allow microphone access when prompted
  2. Speak into your microphone — transcription appears in real-time
  3. Click Stop Recording when finished
  4. Use Clear to reset the transcript

Understanding the Metrics

MetricDescription
RTFxReal-Time Factor — how many times faster than real-time. Higher is better (e.g., 10x means 1 second of audio processed in 0.1 seconds)
LatencyTime from sending audio to receiving transcription
TTFWTime To First Word — delay from speech start to first transcription
E2EEnd-to-End latency — total time from speech start to final transcription
Disconnected
Press Space to toggle recording
Audio: 16kHz mono PCM16