Audio Transcription
audio_transcriptionConvert audio files or URLs into text with Whisper models, adding optional timestamps. The tool polls until transcription completes and returns the result inline or as a download URL.
Instructions
Transcribe audio file to text using Whisper models.
Converts audio files to text transcription with optional timestamps. Automatically polls until transcription is complete.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"audio": "<audio>",
"include_ts": "<include_ts>"
},
"name": "audio_transcription"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| lang | No | Language of the recording as an ISO code, e.g. 'en', 'pl'. Set it whenever you know the language: auto-detection can pick the wrong one, e.g. on short or accented audio. It forces the output language, so a wrong code returns a translation instead of a transcript. Auto-detected when unset | |
| audio | Yes | Audio file (base64 encoded or URL) | |
| model | No | Whisper model from the model list. Default WhisperLargeV3Ct2: honours lang, returns segments, supports diarize and word timestamps | WhisperLargeV3Ct2 |
| diarize | No | Label who is speaking; implies include_ts. Only on models marked 'diarize' in the model list (WhisperLargeV3 rejects it). Costs 1.5x, not stacked with a finer ts_level | |
| ts_level | No | Timestamp granularity: 'segment', 'word' or 'char'; implies include_ts. Only the levels a model lists under 'ts_level' in the model list are accepted (WhisperLargeV3: segment only). Finer than segment costs 1.5x on models that price it | |
| include_ts | Yes | Include timestamps in transcription | |
| webhook_url | No | HTTPS URL for deAPI to call when this job finishes. Notifies your own systems; the tool still returns the result here | |
| webhook_secret | No | 32-255 char key deAPI signs the callback with (HMAC-SHA256). Needs webhook_url. A secret given in chat stays in the transcript | |
| return_result_in_response | No | Return transcription inline. Set to False for large files to get download URL instead |
Output Schema
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||