Advanced Style Control

VoiSona Talk allows you to fine-tune the generated speech using various global parameters and style weights (emotions).

Global Parameters

You can adjust the following parameters in the global_parameters object:

Parameter	Description	Range	Default
alp	Age-like parameter (frequency warping).	-1.0 to 1.0	0.0
huskiness	Control the huskiness of the voice.	-20.0 to 20.0	0.0
intonation	Scale factor for pitch contour variation.	0.0 to 2.0	1.0
pitch	Pitch shift in cents (1 cent = 0.01 semitones).	-600 to 600	0.0
speed	Speech rate (speed multiplier).	0.2 to 5.0	1.0
volume	Amplitude multiplier in decibels (dB).	-8.0 to 8.0	0.0
style_weights	Array of style weight coefficients.	Sum typically 1.0	Voice default

Managing Styles (Emotions)

Instead of manually managing the style_weights array (which depends on the internal order of the voice library), you can use the getStyleWeights helper.

1. Fetch Voice Information

Styles differ between voice libraries. First, retrieve the detailed information for your chosen voice.

const voiceInfo = await client.getVoiceInformation('voice_name', '1.0.0');
console.log(voiceInfo.style_names); // e.g., ["Normal", "Happy", "Sad", "Angry"]

2. Map Styles by Name

Use getStyleWeights to create the weight array using intuitive names.

const styleWeights = client.getStyleWeights(voiceInfo, {
  Happy: 0.8,
  Normal: 0.2,
});

Complete Example

import { VoisonaClient } from 'voisona-talk-api';

const client = new VoisonaClient({ /* config */ });

async function run() {
  // 1. Get voice details
  const voiceInfo = await client.getVoiceInformation('tanaka-san_ja_JP', '2.0.1');

  // 2. Configure styles and parameters
  const result = await client.synthesizeAndWait({
    text: 'わーい！今日はとてもいい天気ですね！',
    language: 'ja_JP',
    voice_name: 'tanaka-san_ja_JP',
    global_parameters: {
      style_weights: client.getStyleWeights(voiceInfo, { Happy: 1.0 }),
      speed: 1.2,      // Slightly faster
      pitch: 100,      // Higher pitch
      intonation: 1.5, // More expressive
    },
  });

  console.log(`Audio saved at: ${result.output_file_path}`);
}

Parameter Validation

This library automatically validates these parameters locally before sending the request to VoiSona Talk, providing clear error messages if values are out of range.