Qwen-Audio-3.0-TTS-Plus Voice and Parameter Walkthrough

Using the built-in voices and controllable parameters of qwen-audio-3.0-tts-plus as a guide, a Chinese narration sample is generated to demonstrate how this Plus mode actually sounds in narration, podcast, and product voice announcement scenarios. The piece uses only the input parameters that genuinely exist in the author's route schema: text and voice are required, while the rest, such as rate, pitch, volume, sample_rate, format, seed, instruction, and language_hints, take values according to the schema's enums and defaults. The audio is output as mp3 at 22050 Hz, with speaking rate and pitch controlled through structured instructions, making it easy for developers to reproduce the same synthesis against the submit input snapshot. According to third-party evaluations, the model scores an Elo of 1,236 on Speech Arena (Provider Voices), positioning it toward professional audio quality, but its throughput is about 16 characters/second, making it suitable for scenarios sensitive to naturalness rather than high throughput.