> This page is for Voice APIs.

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://dev.hume.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://dev.hume.ai/_mcp/server.

# Interruption

**Interruption settings control what happens when a user speaks while EVI is responding.** Configure these
settings through the optional `interruption` field on an EVI config. All fields fall back to defaults when
omitted.

These settings build on EVI's [interruptibility](/docs/speech-to-speech-evi/features/interruptibility) feature.
By default, EVI stops responding as soon as it detects speech from the user. Tuning interruption lets you require
more sustained speech before EVI yields, reducing false interruptions from background noise or brief hesitations.

#### [API Reference](/reference/speech-to-speech-evi/configs/create-config#request.body.interruption)

See the full API Reference for interruption.

## Supported parameters

### Minimum interruption duration

**`min_interruption_ms`**: How long the user must speak while EVI is responding before EVI stops and yields the
floor.

* Lower values make EVI more responsive to the user speaking up but increase the chance of noise or backchannels
  stopping EVI unnecessarily.
* Higher values let EVI continue through brief sounds or hesitations, allowing it to finish more of its response
  before yielding.

**Default**: 800ms | **Range**: 50ms to 2,000ms

## Example configurations

### Resilient responses

Prevent brief sounds or hesitations from cutting off EVI mid-response. Useful when users may produce backchannels
like "uh huh" or think aloud while EVI is speaking.

```json
{
  "interruption": {
    "min_interruption_ms": 1500
  }
}
```

> **Tip**
>
> For a fully patient interaction, pair this with a higher `end_of_turn_silence_ms` in the
> [turn detection](/docs/speech-to-speech-evi/configuration/turn-detection) config so EVI also waits longer before
> responding.

### Sensitive interruption

Make EVI yield quickly when the user speaks up, prioritizing responsiveness over resilience to noise.

```json
{
  "interruption": {
    "min_interruption_ms": 100
  }
}
```

---