Qwen3-TTS Studio

Test a preset voice

Pick one of the built-in CustomVoice speakers and synthesize any text. Add a tone instruction to steer emotion or delivery.

Clone Lab

Upload a clean reference clip and its exact transcript, then test before saving. Tweak the clip, transcript or quality mode and re-test until it sounds right.

Create a voice from a description

Describe the voice you want and VoiceDesign invents it. We render a reference clip and lock it in, so the voice stays consistent every time you use it.

Your voices

Saved presets, clones and designed voices. Type text and click Speak to synthesize with any of them.

REST API

Generate speech programmatically. All endpoints return audio/wav.

POST /api/tts — main generation endpoint

POST /api/tts with a saved voice from your library:

Other endpoints:

GET    /api/meta                 speakers, languages, device
GET    /api/voices               list saved voices
POST   /api/voices/clone         multipart: name, ref_text, language, file
POST   /api/voices/design        json: name, instruct, language, sample_text
POST   /api/voices/preset        json: name, speaker, language, instruct
POST   /api/voices/{id}/speak    json: text, language -> audio/wav
GET    /api/voices/{id}/ref      reference audio for a clone/design
DELETE /api/voices/{id}          remove a voice