Voices

A voice determines the speaker's timbre. You select one by its voice id; if you omit it, the model's default voice is used.

System voices

The TTS model ships with a large catalog of system voices spanning many languages and styles — Mandarin, Cantonese, English, Japanese, Korean, Spanish, Portuguese, French, and more, including character and personality voices.

A voice id looks like male-qn-qingse, English_Trustworthy_Man, or Chinese (Mandarin)_Warm_Girl. Voice ids are case-sensitive and must match exactly.

You can browse and preview the full list in the PlayKit Dashboard voice picker for the TTS model.

{ "voice": "English_Trustworthy_Man" }

If the voice id is not in the catalog, the request is rejected — so prefer copying ids from the dashboard.

Voice mixing

You can blend up to 4 voices into a single custom timbre by giving each a weight from 1–100. Higher weight means more of that voice in the result. Voice mixing is mutually exclusive with picking a single voice.

{
  "voice_mix": [
    { "voice": "female-chengshu", "weight": 70 },
    { "voice": "female-tianmei", "weight": 30 }
  ]
}

Voice tuning

Independently of which voice you use, you can adjust speed, volume, and pitch (see your SDK's voice_settings / equivalent), and set an emotion — covered in Tone & Markup.

For the exact call in your language, see the JavaScript or Unity TTS guide.