# Soniox

> Strong on Hindi–English mixing. Every listening setting for Soniox, and how to set it.

Source: https://voxio.in/docs/voice-listening-soniox · voxioagent 2.0.0 · Index: https://voxio.in/docs/llms.txt

Service: `soniox`. A new setup starts from:

```json
{
  "service": "soniox",
  "model": "stt-rt-v4",
  "language": [
    "hi",
    "en"
  ]
}
```

## Set it

On the flow's **Voice** tab, choose **Listening**, then **Soniox**. Or from code:

```js
await voxio.flows.update(flowKey, {
  stt: {"service":"soniox","model":"stt-rt-v4","language":["hi","en"]},
});
```

## Settings

Settings marked “under More settings” are behind **More settings** in the dashboard.

- `model` (string): **Model**. Which speech model listens. Hover an option to compare. Default: `"stt-rt-v4"`.
- `language` (string[]): **Languages**. Languages callers are likely to speak.
- `language-hints-strict` (boolean): **Only these languages** (under More settings). Ignore any language not listed. Default: `false`.
- `keyterm` (string[]): **Boost words**. Words the agent should never mishear.
- `context-text` (string): **Context** (under More settings). Background that helps recognition.
- `enable-endpoint-detection` (boolean): **Detect end of speech** (under More settings). Let the model decide when a sentence has ended. Default: `true`.
- `max-endpoint-delay-ms` (number (ms)): **Max end-of-speech delay** (under More settings). Longest wait for the end of a sentence. From 500 to 3000. Default: `2000`.
- `endpoint-sensitivity` (number): **End-of-speech sensitivity** (under More settings). Higher ends sentences sooner. From -1 to 1. Default: `0`. Only applies to some models.
- `endpoint-latency-adjustment-level` (number): **Latency adjustment** (under More settings). Trade accuracy for speed. 0 is the most accurate. From 0 to 3. Default: `0`. Only applies to some models.
- `enable-speaker-diarization` (boolean): **Tell speakers apart** (under More settings). Label who said what when more than one person talks. Default: `false`.
- `enable-language-identification` (boolean): **Identify language** (under More settings). Detect which language each sentence is in. Default: `false`.

## Choices

### Model choices

| Value | Name | Notes |
| --- | --- | --- |
| `stt-rt-v5` | Real-time v5 | Latest. Newest Soniox model, with finer control over when a sentence ends. |
| `stt-rt-v4` | Real-time v4 | Proven on Hindi-English mixing. Our current default. |

## Next steps

- [Listening](https://voxio.in/docs/voice-listening.md): compare the other providers.
- [Turn-taking](https://voxio.in/docs/turn-taking.md): when the agent answers.