> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/QwenLM/Qwen3-TTS/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog

> Release history and updates for Qwen3-TTS

## Latest Release

### January 22, 2026 - Qwen3-TTS Initial Release

<img src="https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen3-TTS-Repo/qwen3_tts_logo.png" width="400" />

We are excited to announce the initial release of **Qwen3-TTS**, a series of powerful speech generation models developed by the Qwen team.

#### Released Models

All models are based on the **Qwen3-TTS-Tokenizer-12Hz**:

<CardGroup cols={2}>
  <Card title="Qwen3-TTS-12Hz-1.7B-CustomVoice" icon="microphone">
    1.7B parameter model with 9 premium speakers and instruction control
  </Card>

  <Card title="Qwen3-TTS-12Hz-1.7B-VoiceDesign" icon="wand-magic-sparkles">
    1.7B parameter model with voice design from natural language descriptions
  </Card>

  <Card title="Qwen3-TTS-12Hz-1.7B-Base" icon="clone">
    1.7B parameter base model with 3-second voice cloning capability
  </Card>

  <Card title="Qwen3-TTS-12Hz-0.6B-CustomVoice" icon="microphone">
    0.6B parameter efficient model with 9 premium speakers
  </Card>

  <Card title="Qwen3-TTS-12Hz-0.6B-Base" icon="clone">
    0.6B parameter efficient base model with voice cloning
  </Card>

  <Card title="Qwen3-TTS-Tokenizer-12Hz" icon="code">
    High-fidelity 12Hz speech tokenizer for encoding and decoding
  </Card>
</CardGroup>

#### Key Features

<AccordionGroup>
  <Accordion title="Powerful Speech Representation" icon="signal">
    * Self-developed Qwen3-TTS-Tokenizer-12Hz
    * Efficient acoustic compression and high-dimensional semantic modeling
    * Preserves paralinguistic information and acoustic environmental features
    * High-speed, high-fidelity speech reconstruction
  </Accordion>

  <Accordion title="Universal End-to-End Architecture" icon="diagram-project">
    * Discrete multi-codebook LM architecture
    * Full-information end-to-end speech modeling
    * Eliminates information bottlenecks of traditional LM+DiT schemes
    * Enhanced versatility, generation efficiency, and performance ceiling
  </Accordion>

  <Accordion title="Extreme Low-Latency Streaming" icon="bolt">
    * Dual-Track hybrid streaming generation architecture
    * Single model supports both streaming and non-streaming
    * First audio packet after single character input
    * End-to-end synthesis latency as low as **97ms**
  </Accordion>

  <Accordion title="Intelligent Voice Control" icon="brain">
    * Natural language instruction-based voice control
    * Flexible control over timbre, emotion, and prosody
    * Deep text semantic understanding
    * Adaptive tone, rhythm, and emotional expression
  </Accordion>

  <Accordion title="Comprehensive Language Support" icon="globe">
    Supports 10 major languages:

    * Chinese (Mandarin)
    * English
    * Japanese
    * Korean
    * German
    * French
    * Russian
    * Portuguese
    * Spanish
    * Italian

    Plus Chinese dialects: Beijing, Sichuan
  </Accordion>
</AccordionGroup>

#### Premium Speakers

CustomVoice models include 9 carefully crafted speakers:

| Speaker       | Voice Description                               | Native Language   |
| ------------- | ----------------------------------------------- | ----------------- |
| **Vivian**    | Bright, slightly edgy young female voice        | Chinese           |
| **Serena**    | Warm, gentle young female voice                 | Chinese           |
| **Uncle\_Fu** | Seasoned male voice with low, mellow timbre     | Chinese           |
| **Dylan**     | Youthful Beijing male voice, clear and natural  | Chinese (Beijing) |
| **Eric**      | Lively Chengdu male voice, husky brightness     | Chinese (Sichuan) |
| **Ryan**      | Dynamic male voice with strong rhythmic drive   | English           |
| **Aiden**     | Sunny American male voice, clear midrange       | English           |
| **Ono\_Anna** | Playful Japanese female voice, light and nimble | Japanese          |
| **Sohee**     | Warm Korean female voice with rich emotion      | Korean            |

#### Performance Highlights

* **Best-in-class English synthesis**: WER of 1.24 on Seed-TTS test-en
* **Competitive Chinese synthesis**: WER of 0.77 on Seed-TTS test-zh
* **State-of-the-art voice design**: Leading performance on InstructTTSEval
* **Strong cross-lingual**: Excellent results on language transfer tasks
* **Multilingual excellence**: Competitive performance across all 10 languages

See the [Benchmarks](/resources/benchmarks) page for detailed results.

#### Resources

<CardGroup cols={2}>
  <Card title="Technical Paper" icon="file-lines" href="https://arxiv.org/abs/2601.15621">
    Read the research paper on arXiv
  </Card>

  <Card title="Blog Post" icon="blog" href="https://qwen.ai/blog?id=qwen3tts-0115">
    Detailed introduction and use cases
  </Card>

  <Card title="Hugging Face" icon="huggingface" href="https://huggingface.co/collections/Qwen/qwen3-tts">
    Download models from Hugging Face
  </Card>

  <Card title="ModelScope" icon="cube" href="https://modelscope.cn/collections/Qwen/Qwen3-TTS">
    Download models from ModelScope (CN)
  </Card>

  <Card title="GitHub Repository" icon="github" href="https://github.com/QwenLM/Qwen3-TTS">
    Source code and examples
  </Card>

  <Card title="Live Demo (HF)" icon="play" href="https://huggingface.co/spaces/Qwen/Qwen3-TTS">
    Try Qwen3-TTS in your browser
  </Card>
</CardGroup>

#### Installation

```bash theme={null}
# Install via PyPI
pip install qwen-tts

# Or install from source
git clone https://github.com/QwenLM/Qwen3-TTS.git
cd Qwen3-TTS
pip install -e .

# Optional: Install FlashAttention for better performance
pip install flash-attn --no-build-isolation
```

#### Quick Start

```python theme={null}
import torch
import soundfile as sf
from qwen_tts import Qwen3TTSModel

# Load CustomVoice model
model = Qwen3TTSModel.from_pretrained(
    "Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice",
    device_map="cuda:0",
    dtype=torch.bfloat16,
    attn_implementation="flash_attention_2",
)

# Generate speech
wavs, sr = model.generate_custom_voice(
    text="Hello! Welcome to Qwen3-TTS.",
    language="English",
    speaker="Ryan",
    instruct="Say it with enthusiasm",
)

sf.write("output.wav", wavs[0], sr)
```

#### What's Coming

<Note>
  Additional models mentioned in the technical report will be released in the near future. Stay tuned!
</Note>

***

## Version History

### qwen-tts Package

#### v0.1.1 - Current Release

**Package Information:**

* Python support: 3.9, 3.10, 3.11, 3.12, 3.13
* License: Apache-2.0
* Command-line tool: `qwen-tts-demo`

**Dependencies:**

* transformers 4.57.3
* accelerate 1.12.0
* gradio (latest)
* librosa, torchaudio, soundfile
* onnxruntime, einops

***

## Related Releases

### DashScope API

Production-ready API access for Qwen3-TTS models:

<CardGroup cols={3}>
  <Card title="CustomVoice API" icon="microphone" href="https://help.aliyun.com/zh/model-studio/qwen-tts-realtime">
    Real-time streaming API
  </Card>

  <Card title="Voice Clone API" icon="clone" href="https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning">
    Voice cloning API
  </Card>

  <Card title="Voice Design API" icon="wand-magic-sparkles" href="https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design">
    Voice design API
  </Card>
</CardGroup>

### vLLM-Omni Support

vLLM officially provides day-0 support for Qwen3-TTS:

* Optimized inference engine
* Better throughput and latency
* Offline inference (online serving coming soon)
* See [vLLM-Omni documentation](https://docs.vllm.ai/projects/vllm-omni/en/latest/getting_started/quickstart/)

***

## Stay Updated

<CardGroup cols={3}>
  <Card title="GitHub Releases" icon="tag" href="https://github.com/QwenLM/Qwen3-TTS/releases">
    Watch for new releases
  </Card>

  <Card title="Discord Community" icon="discord" href="https://discord.gg/CV4E9rpNSD">
    Join discussions
  </Card>

  <Card title="WeChat Group" icon="weixin" href="https://github.com/QwenLM/Qwen/blob/main/assets/wechat.png">
    Chinese community
  </Card>
</CardGroup>

<Note>
  Subscribe to the [GitHub repository](https://github.com/QwenLM/Qwen3-TTS) to get notified about new releases and updates.
</Note>