No voice model selected yet
Choose a generation model
Voice-over language Qwen3 2.5 requires an explicit language
Tone control mode Use original audioBoth timbre and delivery follow the voice model's original audio — the simplest option.Set tone parametersTimbre still follows the original audio while delivery comes from your parameters. Adjust only one or two main emotions — maxing them all out sounds unnatural.Upload emotion referenceTimbre follows the voice model, delivery follows the audio you upload. The closer the reference emotion is to your target, the better.
* Both timbre and delivery follow the voice model's original audio — the simplest option.
Speech rate
1.0x