Qwen3-TTS Update! 49 Timbres + 10 Languages + 9 Dialects

DASHSCOPE HUGGING FACE DEMO MODELSCOPE DEMO

Qwen3-TTS-Flash is a flagship text-to-speech model that supports multi-timbre, multi-lingual, and multi-dialect speech synthesis. It aims to produce natural and expressive speech and is available via Qwen API.

Major Improvements:

• Richer Timbres Support: Qwen3-TTS offers over 49 high-quality timbres, covering a range of genders, ages, regional traits, and character profiles to meet diverse scenario needs. Explore a variety of roles such as playful and quirky Momo, the warm and supportive childhood friend Ono Anna, the proud and forthright “tough girl” Vivian, the strict instructor Elias, the wise elder Eldric Sage, the cute loli Bunny, and many more.

• Enhanced Multilingual and Dialect Capabilities: Qwen3-TTS supports 10 major languages, including Chinese, English, German, Italian, Portuguese, Spanish, Japanese, Korean, French, and Russian. On the MiniMax TTS multilingual test set, the model achieves a lower average word error rate (WER) than MiniMax, ElevenLabs, and GPT-4o-Audio-Preview. It also supports dialect synthesis for more dialects, including Mandarin, Hokkien, Wu, Cantonese, Sichuanese, Beijing, Nanjing, Tianjin, and Shaanxi dialects, authentically reproducing local accents and linguistic nuances.

• More Natural and Human-like Prosody/Speech Rates: Compared to the previous version, Qwen3-TTS has greatly improved its ability to adaptively adjust speech rate and prosody according to textual input, achieving a level of human-likeness that closely approximates real human speech.

Samples

Qwen3-TTS offers a diverse range of distinctive, emotionally rich timbres for users to choose from, meeting the needs of a variety of scenarios. Here are some sample synthesized samples:

Qwen3-TTS provides comprehensive support for a variety of Chinese dialects, accurately capturing regional accents, tones, and local charm. Here are some sample synthesized samples:

Qwen3-TTS also supports authentic and natural multilingual timbres, with speech patterns that closely resemble native pronunciation. Here are some sample synthesized samples:

Performance

How to use

Use Qwen3-TTS with Qwen API is simple. We demonstrate a code snippet for you to play with it below:

# Please install the latest version of the DashScope SDK.
    import os
    import requests
    import dashscope
    
    text = "Let me recommend a T-shirt to everyone. This one is really super good-looking, and the color is very classy. It’s also a great piece to mix and match with anything, so you can totally buy it without hesitation. It looks amazing and is very forgiving on the figure—no matter what body type you have, it will look great on you. Highly recommend you place an order!"
    # Usage of the SpeechSynthesizer Interface: dashscope.audio.qwen_tts.SpeechSynthesizer.call(...)
    response = dashscope.MultiModalConversation.call(
        model="qwen3-tts-flash-2025-11-27",
        api_key=os.getenv("DASHSCOPE_API_KEY"),
        text=text,
        voice="Ryan",
        language_type="English", # It is recommended to match the language with the text in order to obtain correct pronunciation and natural intonation.
        stream=False
    )
    audio_url = response.output.audio.url
    save_path = "downloaded_audio.wav"  # Custom save path.
    
    try:
        response = requests.get(audio_url)
        response.raise_for_status()  # Check whether the request was successful.
        with open(save_path, 'wb') as f:
            f.write(response.content)
        print(f"The audio file has been saved to: {save_path}")
    except Exception as e:
        print(f"Download failed: {str(e)}")
    

Citation

If you find our model useful in your research, please consider citing it 📝 :)

@misc{qwen3_tts_202511,
      author       = {Qwen Team, Alibaba},
      title        = {Qwen3-TTS Update! 49 Timbres + 10 Languages + 9 Dialects},
      year         = {2025},
      url          = {https://qwen.ai/blog?id=qwen3-tts-1128},
      urldate      = {2025-12-01}
    }