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

# Speech recognition

<p className="lead">
  Speech recognition resources control how the agent processes user speech input on the voice channel.
</p>

These resources live under `voice/speech_recognition/` and are used to tune how the agent listens, recognizes, and post-processes spoken input.

<Note>
  **ASR settings are platform-provisioned — update only**

  ASR settings are created automatically when a project is created. They can be updated with `poly push` but not created from scratch. See the [equivalent note on agent settings](/adk/reference/agent_settings) for details.
</Note>

## Location

```text theme={"theme":{"light":"github-light","dark":"github-dark"}}
voice/speech_recognition/
├── asr_settings.yaml
├── keyphrase_boosting.yaml
└── transcript_corrections.yaml
```

All three files are voice-specific. Only `asr_settings.yaml` is the core settings file; the others are optional.

## What speech recognition controls

<CardGroup>
  <Card title="ASR settings">
    Configure global speech-recognition behavior such as barge-in and latency/accuracy style.
  </Card>

  <Card title="Keyphrase boosting">
    Bias recognition toward specific words or phrases.
  </Card>

  <Card title="Transcript corrections">
    Apply regex-based corrections after speech recognition.
  </Card>
</CardGroup>

## ASR settings

ASR settings are defined in:

```text theme={"theme":{"light":"github-light","dark":"github-dark"}}
voice/speech_recognition/asr_settings.yaml
```

These settings control global speech-recognition behavior for the voice channel.

### Fields

| Field               | Type     | Description                                                                      |
| ------------------- | -------- | -------------------------------------------------------------------------------- |
| `barge_in`          | `bool`   | Whether the user can interrupt the agent while it is speaking. Default: `false`. |
| `interaction_style` | `string` | Controls the latency/accuracy trade-off. Default: `balanced`.                    |

### Example

```yaml theme={"theme":{"light":"github-light","dark":"github-dark"}}
barge_in: false
interaction_style: balanced
```

### Interaction styles

| Style             | Behavior                                  |
| ----------------- | ----------------------------------------- |
| `precise`         | Higher accuracy, higher latency           |
| `balanced`        | Default balance of speed and accuracy     |
| `swift`           | Faster responses, slightly lower accuracy |
| `sonic` / `turbo` | Lowest latency                            |

## Keyphrase boosting

Keyphrase boosting is defined in:

```text theme={"theme":{"light":"github-light","dark":"github-dark"}}
voice/speech_recognition/keyphrase_boosting.yaml
```

It biases the recognizer toward specific words or phrases, which is useful for:

* brand names
* product names
* specialist terminology
* domain-specific jargon

### Structure

A `keyphrases` list where each entry includes:

| Field       | Required | Description                                        |
| ----------- | -------- | -------------------------------------------------- |
| `keyphrase` | Yes      | The word or phrase to boost                        |
| `level`     | No       | Boost strength: `default`, `boosted`, or `maximum` |

### Example

```yaml theme={"theme":{"light":"github-light","dark":"github-dark"}}
keyphrases:
  - keyphrase: PolyAI
    level: maximum
  - keyphrase: reservation
    level: boosted
  - keyphrase: check-in
    level: default
```

## Transcript corrections

Transcript corrections are defined in:

```text theme={"theme":{"light":"github-light","dark":"github-dark"}}
voice/speech_recognition/transcript_corrections.yaml
```

These rules post-process ASR output to fix common misrecognitions.

They are especially useful for:

* email domains
* repeated digits
* domain-specific phrases
* spoken forms that should be normalized into machine-friendly text

### Structure

A `corrections` list where each entry includes:

| Field                 | Required | Description                         |
| --------------------- | -------- | ----------------------------------- |
| `name`                | Yes      | Identifier for the correction group |
| `description`         | No       | Explains what the correction fixes  |
| `regular_expressions` | Yes      | Regex rules used for correction     |

Each regex rule can include:

| Field                | Required | Description                       |
| -------------------- | -------- | --------------------------------- |
| `regular_expression` | Yes      | Pattern to match                  |
| `replacement`        | Yes      | Replacement text                  |
| `replacement_type`   | No       | `full` or `partial` / `substring` |

### Example

```yaml theme={"theme":{"light":"github-light","dark":"github-dark"}}
corrections:
  - name: Email domain fix
    description: Correct common email domain misrecognitions
    regular_expressions:
      - regular_expression: at gmail dot com
        replacement: "@gmail.com"
        replacement_type: full
      - regular_expression: at hotmail dot com
        replacement: "@hotmail.com"
        replacement_type: full

  - name: Number normalization
    description: Normalize spoken numbers to digits
    regular_expressions:
      - regular_expression: \bdouble (\d)\b
        replacement: \1\1
        replacement_type: partial
```

## Best practices

* use `keyphrase_boosting` for terms the recognizer is likely to miss
* keep boosted keyphrases focused and specific
* use transcript corrections for common, repeated recognition errors
* avoid overly broad regex rules that may alter normal input unexpectedly
* choose the ASR interaction style deliberately based on latency and accuracy needs

<Tip>
  **Use the lightest possible intervention**

  Start with the default settings, then add boosting or transcript corrections only where recognition problems are actually recurring.
</Tip>

## Related pages

<CardGroup>
  <Card title="Voice settings" href="/adk/reference/voice_settings">
    See how speech recognition fits into the wider voice-channel configuration.
  </Card>

  <Card title="Response control" href="/adk/reference/response_control">
    Configure what happens to output before it is spoken.
  </Card>
</CardGroup>