> ## 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.

# Call Model — STT on edge

> Whisper.cpp and Faster Whisper catalog models, dependencies, and compile checks

**Call Model** (`call_model`) nodes that select an **edge** or **hybrid** speech-to-text catalog model run inference on-device via `cw.models.load(...).predict(...)`. Cloud-only STT uses `client.mlmodels.run(...)` and does not require edge STT packages.

<Note>
  Prompting for STT models is optional biasing — see [Prompting Call Model](/feature-reference/workflows/call-model-prompt).
</Note>

***

## Catalog models and dependencies

Each STT seed sets **one** SDK extra in `metadata.edge_package`. Edge-sync surfaces it in `model_requirements`.

| Model                           | Runtime          | Edge extra      | Install                                  |
| ------------------------------- | ---------------- | --------------- | ---------------------------------------- |
| Whisper Tiny EN Q5\_1           | `whisper_cpp`    | `ml-stt`        | `pip install 'cyberwave[ml-stt]'`        |
| Whisper Base EN Q5\_1           | `whisper_cpp`    | `ml-stt`        | same                                     |
| Whisper Small EN Q5\_1          | `whisper_cpp`    | `ml-stt`        | same                                     |
| Whisper Tiny Multilingual Q5\_1 | `whisper_cpp`    | `ml-stt`        | same (+ cloud fallback)                  |
| Whisper Base Multilingual Q5\_1 | `whisper_cpp`    | `ml-stt`        | same (+ cloud fallback)                  |
| Faster Whisper Tiny EN          | `faster_whisper` | `ml-stt-faster` | `pip install 'cyberwave[ml-stt-faster]'` |
| Faster Whisper Base EN          | `faster_whisper` | `ml-stt-faster` | same                                     |
| Faster Whisper Small EN         | `faster_whisper` | `ml-stt-faster` | same                                     |

Do **not** point catalog metadata at `ml-all`. Pick the extra for the runtime you use.

***

## Weight paths (`edge_model_path`)

| Model family       | Cache / load path                                                                                  | Notes                                                                                                   |
| ------------------ | -------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------- |
| **whisper.cpp**    | File path to GGML binary, e.g. `models/whisper/ggml-tiny.en-q5_1.bin`                              | `download_url` in metadata; first run fetches into edge model cache                                     |
| **faster-whisper** | Cache key = `model_external_id` (e.g. `tiny.en`); `metadata.edge_model_path` is documentation only | Edge Core pre-downloads via catalog `model_external_id`; CTranslate2 cache under `/app/models/tiny.en/` |

Generated workers emit:

```python theme={null}
model = cw.models.load(
    "tiny.en",
    runtime="faster_whisper",
    faster_whisper_model_id="tiny.en",
    download_root="tiny.en",
    ...
)
```

***

## Compile server

When a workflow has `run_on_edge: true` and a Call Model node references an on-device STT model, compile verifies imports:

| Runtime          | Import checked                | `base.txt` package      |
| ---------------- | ----------------------------- | ----------------------- |
| `whisper_cpp`    | `pywhispercpp.model.Model`    | `pywhispercpp==1.4.1`   |
| `faster_whisper` | `faster_whisper.WhisperModel` | `faster-whisper>=1.1.1` |

Failure example:

> This workflow uses Call Model with faster-whisper but faster-whisper is not installed on the compile server. Install faster-whisper (or cyberwave\[ml-stt-faster]) on Django before compiling for edge.

Rebuild Django after updating `requirements/base.txt` — see [Edge workflow dependencies](/feature-reference/workflows/edge-dependencies).

***

## Edge worker

1. Install the catalog model's extra (`ml-stt` or `ml-stt-faster`).
2. Sync workflow — `edge-sync` lists `model_requirements` with `edge_model_path`.
3. First inference downloads weights (or uses pre-staged files under `~/.cyberwave/models/`).

**Audio input:** int16 PCM @ 16 kHz mono (or WAV bytes). The worker passes `sample_rate_hz`, `channels`, and optional `language` / `task` / `vad_filter` (Faster Whisper built-in VAD when enabled).

***

## Choosing whisper.cpp vs Faster Whisper

|                  | whisper.cpp (`ml-stt`)            | Faster Whisper (`ml-stt-faster`)      |
| ---------------- | --------------------------------- | ------------------------------------- |
| **Hardware**     | Raspberry Pi 4 class              | Pi 4+; Jetson/GPU for Base/Small      |
| **Weights**      | Quantized GGML (`.bin`)           | CTranslate2 cache dir                 |
| **Latency**      | Good for Tiny on CPU              | Tiny EN optimized for real-time       |
| **Built-in VAD** | No (use Audio Assistant upstream) | Yes (`builtin_vad_filter` on catalog) |

***

## Typical wiring

```
audio_track → audio_assistant → wake_word_engine (optional)
  → call_model (STT catalog model)
  → fuzzy_matcher ← twin
  → virtual_controller
```

Wire Call Model **audio** input from upstream `audio` key. Wire **result** (text) into Fuzzy Matcher's **Uncertain String** (`query`).

***

## Related

<CardGroup cols={2}>
  <Card title="Edge dependencies" icon="puzzle-piece" href="/feature-reference/workflows/edge-dependencies">
    Full matrix (all nodes + compile server)
  </Card>

  <Card title="Audio in Workflows" icon="waveform-lines" href="/feature-reference/workflows/audio-in-workflows">
    PCM format and pipelines
  </Card>

  <Card title="Fuzzy Matcher" icon="bullseye" href="/feature-reference/workflows/fuzzy-matcher">
    Map STT text to commands
  </Card>
</CardGroup>