Qwen-Audio-3.0-TTS lands with real-time speech and richer voice control
通义实验室 · wechat · 2026-07-20
Alibaba Tongyi Lab has released Qwen-Audio-3.0-TTS, a real-time text-to-speech model with two variants:
- Flash: optimized for live interaction, with roughly 300ms first-audio latency.
- Plus: optimized for higher-quality synthesis, with better naturalness and voice similarity.
The update focuses on four production pain points: broader language coverage, more natural instruction following, finer-grained tag control, and robustness when the reference audio is noisy or unclear. The release claims top rank on ArtificialAnalysis for Qwen-Audio-3.0-TTS-Plus.
It supports 16 languages and 20 high-quality Chinese dialects, plus natural-language control over emotion,角色/scene/style/speaking rate. It also supports embedded tags such as [gasp], [giggles], and [angry] for fine-grained prosody control.
Other upgrades include stronger noise/reverb robustness, a curated voice library, 48 kHz output for premium scenarios, and up to 3 minutes per synthesis request.
Related event: Alibaba Releases Qwen-Audio-3.0-TTS Voice Synthesis Model(5 posts)→
More from Multimodal
- FastH3-Live hits 22fps: acceleration node benchmarks and the --vram-headroom trick — spartong945 · 2026-09-11
- Midjourney style code share: --sref 2912175708 — tisch_eins · 2026-09-11
- Astra storyboards plus Minimax H3 per-shot generation boost video success rates — Hailuo_AI · 2026-09-11
- MiniMax H3 MAX nails cooking anime clips: 15-second curry demo with prompts shared — Hailuo_AI · 2026-09-11
- MiniMax Music Production Toolkit 2.5 for ComfyUI adds full mastering chain — Vivid_Promise1700 · 2026-09-11
- New Node Finder for ComfyUI ranks fresh nodes by star velocity and recency — Luke2642 · 2026-09-11