Skip to main content
POST
/
audio
/
transcriptions
/api/v1/audio/transcriptions
curl --request POST \
  --url https://api.venice.ai/api/v1/audio/transcriptions \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: multipart/form-data' \
  --form file='@example-file' \
  --form model=nvidia/parakeet-tdt-0.6b-v3 \
  --form response_format=json \
  --form timestamps=false
{
  "text": "<string>",
  "duration": 123,
  "timestamps": {
    "word": [
      {
        "word": "<string>",
        "start": 123,
        "end": 123
      }
    ],
    "segment": [
      {
        "text": "<string>",
        "start": 123,
        "end": 123
      }
    ],
    "char": [
      {
        "char": "<string>",
        "start": 123,
        "end": 123
      }
    ]
  }
}

Authorizations

​
Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

multipart/form-data

Request to transcribe audio to text.

​
file
file

The audio file object (not a base64 string). Supported formats: WAV, WAVE, FLAC, M4A, AAC, MP4, MP3, OGG, WEBM.

​
model
enum<string>
default:nvidia/parakeet-tdt-0.6b-v3

The model to use for transcription. See https://docs.venice.ai/models/overview for more information.

Available options:
nvidia/parakeet-tdt-0.6b-v3,
openai/whisper-large-v3
Example:

"nvidia/parakeet-tdt-0.6b-v3"

​
response_format
enum<string>
default:json

The format of the transcript output, in one of these options: json, text.

Available options:
json,
text
Example:

"json"

​
timestamps
boolean
default:false

Whether to include timestamps in the response.

Example:

false

​
language
string

ISO 639-1 language code (e.g., "en", "es", "fr"). Optional - if not provided, the model will auto-detect the language. Note: Only supported by certain models (e.g., Whisper). Ignored by models that do not support language hints.

Example:

"en"

Response

Transcription completed successfully

Transcription response

​
text
string
required

The transcribed text

​
duration
number

Duration of the audio in seconds

​
timestamps
object

Timestamps for the transcription (only if timestamps=true)