Speech-to-Text REST API audio formats

Applies to both file (multipart upload) and file_url (URL download).

Common

Max size: 100 MB
Processing:
- Whisper, distil-whisper, Parakeet, and Omi models: audio is decoded, resampled to 16kHz, and mixed to mono via ffmpeg before transcription. Container format doesn’t matter as long as ffmpeg can decode it — the validated extension list is the actual restriction.
- deepgram/* models: the file is forwarded to Deepgram as-is, without decoding or resampling, so only mp3 and wav are accepted. Deepgram mixes channels down and transcribes once; with model_config.multichannel, each channel is transcribed separately — see Multichannel.

Format	`whisper-turbo`	`whisper-tiny`	`deepgram/nova-3`	Parakeet (`nvidia/parakeet-v3`, `omi-health/omi-med-stt-v1`)
flac	Yes	Yes	No	Yes
m4a	Yes	Yes	No	Yes
mp3	Yes	Yes	Yes	Yes
mp4	Yes	Yes	No	Yes
mpeg	Yes	Yes	No	Yes
mpga	Yes	Yes	No	Yes
oga	Yes	Yes	No	Yes
ogg	Yes	Yes	No	Yes
wav	Yes	Yes	Yes	Yes
webm	Yes	Yes	No	Yes

	`file`	`file_url`
Delivery	Multipart upload in request body	Server downloads from URL before transcription
Timeout	Request timeout	15s download timeout
Auth	N/A	URL must be publicly accessible (no auth headers forwarded)
Validation	Same format and size checks	Same format and size checks

One of file or file_url is required. Sending both returns 400.