POST /v1/audio/transcriptions

A multipart form upload, compatible with OpenAI's endpoint of the same name.

Field	Required	Meaning
file	yes	The audio: WAV, MP3, M4A, AAC, OGG/Opus, WebM, FLAC, CAF or AIFF. Video files (MP4, MOV, 3GP, MKV, WebM) work too; only the first audio track is used. The type is read from the file's contents, not its name, and other files are refused.
model	no	loquela-remote-beta, or leave it out. For clients that cannot change the model name, whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe and gpt-transcribe are accepted and use the same model. gpt-4o-transcribe-diarize is accepted too and turns on speaker labels.
response_format	no	json (default), text, verbose_json, srt, vtt or diarized_json. See response formats.
diarize	no	true adds speaker labels. Off unless you ask for it.
timestamp_granularities[]	no	With verbose_json: word and/or segment.
language	no	English only. en is accepted; other languages return an error.
prompt, temperature	no	Accepted and ignored. The model does not use them.

Large recordings: uploads are limited to 100 MB, but compressed audio fits far more. About 3 hours of speech fits as a 64 kbps MP3 or Opus file. See limits.

The response carries an x-request-id header. Quote it if you contact us; it is the only thing we can look up, since nothing about the audio is kept.