# Voice

> How the agent hears the caller (speech-to-text) and how it sounds (text-to-speech).

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

A flow has one listening setup, `stt`, and one speaking setup, `tts`. Each is a `service` plus that provider's settings. Change them on the flow's **Voice** tab, save setups to reuse on** Voices**, or try a change for a single call from **Settings for this test** on the Test tab.

```js
await voxio.flows.update(flowKey, {
  stt: { service: "deepgram-streaming", model: "nova-3", language: "en-IN", keyterm: ["Voxio"] },
  tts: { service: "sarvam", model: "bulbul:v3", voice: "simran", language: "en-IN", pace: 1.05 },
});
```

## Pick a setup

| Callers speak | Listening | Speaking |
| --- | --- | --- |
| English, Indian English | Deepgram Nova 3 | Deepgram Aura or Sarvam Bulbul v3 (en-IN) |
| Hindi and English mixed | Soniox | Sarvam Bulbul v3 |
| Other Indian languages | Sarvam Saaras | Sarvam Bulbul v3 |
| A specific brand voice | Any | ElevenLabs or Resemble with your cloned voice |

## In this section

| Page | Covers |
| --- | --- |
| [Listening](https://voxio.in/docs/voice-listening.md) | Deepgram, Soniox, Sarvam |
| [Speaking](https://voxio.in/docs/voice-speaking.md) | Deepgram Aura, Sarvam, ElevenLabs, Google, Resemble |
| [Emotion](https://voxio.in/docs/voice-emotion.md) | Laughs, whispers and tone with ElevenLabs v3. |

## Flow-wide extras

These sit on the flow (Behaviour tab), not inside `stt` or `tts`. From code, send them in `runtime`:

- `voice-isolation` (number): **Background noise suppression**. Removes background noise from the caller’s audio. From 0 to 1. Default: `0`.
- `grain-voice` (boolean): **Natural texture**. Adds a faint grain so the voice sounds less synthetic. Default: `true`.
- `grain-level` (number): **Texture strength**. How strong the texture is. Only used when `grain-voice` is on. From 0 to 0.15. Default: `0.025`.

## Next steps

- [Listening](https://voxio.in/docs/voice-listening.md): speech-to-text providers.
- [Speaking](https://voxio.in/docs/voice-speaking.md): text-to-speech providers.