Skip to main content
Method: client.generateTranscription().

​
Example

const transcript = await client.generateTranscription({
  model: "openai/gpt-4o-transcribe",
  file: "data:audio/mp3;base64,...",
  language: "en",
});

​
Key parameters

  • model (required): Transcription-capable model id.
  • file (required): Audio data (URL, base64, or binary).
  • language: Hint language code (e.g., en).
  • prompt: Optional context to improve accuracy.
  • temperature (0–2): Sampling randomness.
  • response_format: json, text, srt, verbose_json, etc.

​
Returns

AudioTranscriptionResponse 
{
  "text": "Hello world"
}