Qwen3-TTS Technical Report
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912840479145984 |
|---|---|
| author | Hu, Hangrui Zhu, Xinfa He, Ting Guo, Dake Zhang, Bin Wang, Xiong Guo, Zhifang Jiang, Ziyue Hao, Hongkun Guo, Zishan Zhang, Xinyu Zhang, Pei Yang, Baosong Xu, Jin Zhou, Jingren Lin, Junyang |
| author_facet | Hu, Hangrui Zhu, Xinfa He, Ting Guo, Dake Zhang, Bin Wang, Xiong Guo, Zhifang Jiang, Ziyue Hao, Hongkun Guo, Zishan Zhang, Xinyu Zhang, Pei Yang, Baosong Xu, Jin Zhou, Jingren Lin, Junyang |
| contents | In this report, we present the Qwen3-TTS series, a family of advanced multilingual, controllable, robust, and streaming text-to-speech models. Qwen3-TTS supports state-of-the-art 3-second voice cloning and description-based control, allowing both the creation of entirely novel voices and fine-grained manipulation over the output speech. Trained on over 5 million hours of speech data spanning 10 languages, Qwen3-TTS adopts a dual-track LM architecture for real-time synthesis, coupled with two speech tokenizers: 1) Qwen-TTS-Tokenizer-25Hz is a single-codebook codec emphasizing semantic content, which offers seamlessly integration with Qwen-Audio and enables streaming waveform reconstruction via a block-wise DiT. 2) Qwen-TTS-Tokenizer-12Hz achieves extreme bitrate reduction and ultra-low-latency streaming, enabling immediate first-packet emission ($97\,\mathrm{ms}$) through its 12.5 Hz, 16-layer multi-codebook design and a lightweight causal ConvNet. Extensive experiments indicate state-of-the-art performance across diverse objective and subjective benchmark (e.g., TTS multilingual test set, InstructTTSEval, and our long speech test set). To facilitate community research and development, we release both tokenizers and models under the Apache 2.0 license. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_15621 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Qwen3-TTS Technical Report Hu, Hangrui Zhu, Xinfa He, Ting Guo, Dake Zhang, Bin Wang, Xiong Guo, Zhifang Jiang, Ziyue Hao, Hongkun Guo, Zishan Zhang, Xinyu Zhang, Pei Yang, Baosong Xu, Jin Zhou, Jingren Lin, Junyang Sound Computation and Language Audio and Speech Processing In this report, we present the Qwen3-TTS series, a family of advanced multilingual, controllable, robust, and streaming text-to-speech models. Qwen3-TTS supports state-of-the-art 3-second voice cloning and description-based control, allowing both the creation of entirely novel voices and fine-grained manipulation over the output speech. Trained on over 5 million hours of speech data spanning 10 languages, Qwen3-TTS adopts a dual-track LM architecture for real-time synthesis, coupled with two speech tokenizers: 1) Qwen-TTS-Tokenizer-25Hz is a single-codebook codec emphasizing semantic content, which offers seamlessly integration with Qwen-Audio and enables streaming waveform reconstruction via a block-wise DiT. 2) Qwen-TTS-Tokenizer-12Hz achieves extreme bitrate reduction and ultra-low-latency streaming, enabling immediate first-packet emission ($97\,\mathrm{ms}$) through its 12.5 Hz, 16-layer multi-codebook design and a lightweight causal ConvNet. Extensive experiments indicate state-of-the-art performance across diverse objective and subjective benchmark (e.g., TTS multilingual test set, InstructTTSEval, and our long speech test set). To facilitate community research and development, we release both tokenizers and models under the Apache 2.0 license. |
| title | Qwen3-TTS Technical Report |
| topic | Sound Computation and Language Audio and Speech Processing |
| url | https://arxiv.org/abs/2601.15621 |