> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cyberwave.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Audio Track

> Edge microphone trigger — stream PCM chunks into workflows via Zenoh.

The **Audio Track** trigger fires for every buffered chunk from a twin's edge microphone. Audio is delivered over the Zenoh data bus and handled by `@cw.on_audio` in the compiled worker.

## Edge pipeline

```
Microphone driver → Zenoh → @cw.on_audio → Audio Track FIFO → downstream nodes
```

## Buffer presets (downstream pairing)

| Preset                  | Duration      | Pair with                 |
| ----------------------- | ------------- | ------------------------- |
| Voice Assistant (32 ms) | 512 samples   | Audio Assistant (VA)      |
| Wake Word (80 ms)       | 1280 samples  | Wake Word Engine (direct) |
| Speech-To-Text (4 s)    | 64000 samples | SSG, Whisper / STT        |

The compiler rejects incompatible pairings (e.g. 4 s chunks into Wake Word without VAD).

## Deploy

```bash theme={null}
cyberwave workflow compile <uuid>
cyberwave workflow sync --twin-uuid <twin>
```

## Dependencies

```bash theme={null}
pip install "cyberwave[zenoh]"   # required for @cw.on_audio
```

Downstream nodes add `ml-vad`, `ml-aed`, or `ml-wakeword` as needed. See [feature reference](/feature-reference/workflows/audio-track/overview).