Qwen-Audio-3.0-TTS adds finer voice control and broader language coverage
智东西 · wechat · 2026-07-20
Alibaba’s Qwen team has released Qwen-Audio-3.0-TTS, a new text-to-speech model with two variants:
- Flash: optimized for real-time interaction, with first-token latency at the 300 ms level
- Plus: optimized for high-quality generation
The model adds finer control over tone, emotion, breathing, and style through structured tags such as [gasp], [giggles], and [angry], and it also supports free-form natural language instructions for character, mood, scene, and speaking rate.
Other reported upgrades:
- 16 languages supported, including new coverage for Arabic, Vietnamese, Malay, and Filipino
- In benchmark results, the series ranks first in 10 of 16 languages on WER/CER metrics
- Qwen-Audio-3.0-TTS-Plus gets top speaker-similarity scores across all 16 languages
- Support for 20 Chinese dialects, with improved prosody and tone handling
- Better voice cloning under noisy or reverberant reference audio
- 48 kHz output and up to 3 minutes of synthesis per request
Both Plus and Flash are now available on Alibaba Cloud Bailian.
Related event: Alibaba Releases Qwen-Audio-3.0-TTS Voice Synthesis Model(5 posts)→
More from Multimodal
- Runway App Studio Demos: Character Swap and Production-Ready Character Sheets — runwayml · 2026-10-27
- Guy turns his girlfriend's selfie into a playable AI game ad about popping pimples — CurieuxExplorer · 2026-10-27
- Fibo 2 edit called unstable: even a color change alters the edited object — jd3k · 2026-10-27
- AIArtsy AI generates 3D cartoon-style sword fighting animation clip — Jealous-Associate-41 · 2026-10-27
- Seeking AI workflows for high-quality anime comics and colored manga pages — Worldly_Courage395 · 2026-10-27
- Open-source FableCut lets AI agents edit a real video timeline via remote MCP — FastPresence9799 · 2026-10-27