Skip to main content

​
Overview

You can update a persona’s language, voice generation settings, and voice detection settings during a live session — no reconnect required. Updates are sent over the data channel using sendDataMessage().
Updates are best-effort. Where possible they take effect immediately; in some cases they are applied at the start of the next user turn.

​
Method Signature

sendDataMessage(message: string): void

​
Basic Usage

Send a persona_config message containing only the fields you want to change: 
anamClient.sendDataMessage(
  JSON.stringify({
    message_type: "persona_config",
    data: {
      voiceGenerationOptions: { speed: 1.1 },
    },
  }),
);
Include any combination of languageCode, voiceGenerationOptions, and voiceDetectionOptions inside data. Only the fields you send are changed — everything else keeps its current value.

​
What you can update

​
Language

Set languageCode to an ISO 639-1 code: 
anamClient.sendDataMessage(
  JSON.stringify({
    message_type: "persona_config",
    data: { languageCode: "fr" },
  }),
);

​
Voice generation

Adjust speed, volume, emotion, and provider-specific settings. The available fields and their valid ranges depend on your voice provider and model — see Voice Configuration for the full list. 
anamClient.sendDataMessage(
  JSON.stringify({
    message_type: "persona_config",
    data: {
      voiceGenerationOptions: { speed: 1.2, volume: 1.5 },
    },
  }),
);

​
Voice detection

Adjust turn-taking and silence handling. See Voice Detection for all options and ranges. 
anamClient.sendDataMessage(
  JSON.stringify({
    message_type: "persona_config",
    data: {
      voiceDetectionOptions: { endOfSpeechSensitivity: 0.7 },
    },
  }),
);

​
Validation

Updates are validated before they are applied:
  • At least one field must be present — an empty data: {} is ignored.
  • Unknown field names are rejected.
  • Values must be within the valid range for your provider and model.
  • Sending a voice generation field your provider or model doesn’t support is rejected.
If an update is rejected, the session continues and the invalid update is dropped. The reason is delivered as a SERVER_WARNING event: 
import { AnamEvent } from "@anam-ai/js-sdk";

anamClient.addListener(AnamEvent.SERVER_WARNING, (message) => {
  console.warn("Config update rejected:", message);
});

​
Prerequisites

Before sending an update, you must:
  1. Have an active streaming session (call stream() or streamToVideoElement() first).
  2. Have successfully connected to the Anam Engine.

​
Next Steps

Voice Configuration

See every voice generation option and its valid range per provider

Voice Detection

See every turn-taking and silence option

Event Types

Listen for SERVER_WARNING and other session events

User Messages

Send messages and context programmatically
Last modified on June 17, 2026