yao-audio · diff
git:20260807.5c85c24 to git:20260808.6cdccd5
4 added, 15 removed. Audit A to A.
---
name: yao-audio
description: Audio expert. ALWAYS invoke this skill when the user asks to transcribe, recognize, or convert speech/audio to text.
---
# Audio Tools
Use these tools to transcribe audio files to text using speech-to-text models.
## audio_transcribe
Transcribe an audio file to text.
- ### Local file:
```bash
- tai tool audio_transcribe '{"audio_path": "/workspace/meeting.m4a"}'
- ```
-
- ### URL:
- ```bash
- tai tool audio_transcribe '{"audio_path": "https://example.com/podcast.mp3", "language": "en"}'
+ tai tool audio_transcribe --audio_path /path/to/meeting.m4a
```
- ### With a specific STT provider:
```bash
- tai tool audio_transcribe '{"audio_path": "/path/to/recording.wav", "provider": "llm.my-openai:whisper-1"}'
+ tai tool audio_transcribe --audio_path /path/to/recording.wav --language en --provider llm.my-openai:whisper-1
```
| Parameter | Type | Required | Description |
| ---------- | ------ | -------- | ----------------------------------------------------------------- |
- | audio_path | string | yes | File path or URL to the audio file |
+ | audio_path | string | yes | Audio file path. Supported: mp3, m4a, wav, webm, mp4, mpeg, mpga |
| language | string | no | ISO 639-1 language code (e.g. `en`, `zh`, `ja`). Auto-detected if omitted |
| provider | string | no | STT provider connector ID. If omitted, uses the default STT provider |
- Supported formats: mp3, mp4, mpeg, mpga, m4a, wav, webm (Whisper-compatible formats).
-
- Files larger than 25MB are automatically split into chunks and transcribed sequentially. The user does not need to handle splitting manually.
-
## audio_providers
List available speech-to-text providers and models.
### List STT providers (default):
```bash
- tai tool audio_providers '{}'
+ tai tool audio_providers
```
| Parameter | Type | Required | Description |
| ---------- | ------ | -------- | ------------------------------------ |
| capability | string | no | Filter by capability (default: `audio`) |
Returns a list of providers with their available models and connector IDs that can be passed to `audio_transcribe`.
## Constraints
Only use the parameters listed above for each tool. Do not pass unsupported parameters — they will be ignored or cause errors.