Transcription playground
Upload audio or record from your microphone, get text back with timestamps.
Transcript
text
Full JSON response
Request
cURL
About this playground
Audio is sent to /v1/audio/transcriptions — the OpenAI-compatible endpoint — and transcribed by the selected model. Choose verbose_json for segment timestamps, language and duration, or srt / vtt for subtitle files. Billing is per minute of audio, measured from the file.
Tips
- Set the language when you know it: it improves accuracy on short clips and lowers latency. Per-language notes are on the languages pages.
- Files up to 100 MB are accepted; both models run at roughly 4–5% of real time.
- Microphone recordings are captured as webm in your browser and uploaded the same way as a file.