NetEase Youdao releases Confucius4-TTS: open-source 14-language cross-lingual accent-free voice cloning model
Original title:网易发布 Confucius4-TTS:业内首个支持 14 种语言跨语种无口音、且无需参考文本即可完成语音克隆的开源模型,3 秒音频即可克隆音色
AI summary
NetEase Youdao launched the Confucius4-TTS engine of its Ziyue 4.0, claiming it is the industry's first open-source model that supports accent-free cross-lingual voice cloning in 14 languages and requires no reference text. Users need only 3 seconds of audio for zero-shot timbre cloning, with cloned timbre similarity to the original voice exceeding 85% and task accuracy reaching 97%. The model supports 14 languages including Chinese and English and is the first to enable emotion cloning and transfer via audio prompts. It is built on a GPT-style semantic large model, SSL pretrained features with an ECAPA-TDNN speaker encoder, and a Flow Matching framework. It has been fully open-sourced (Apache license), with a 54GB resource package provided for local deployment.
Key Takeaways
- NetEase Youdao launched the Confucius4-TTS engine of its Ziyue 4.
- 0,claiming it is the industry's first open-source model that supports accent-free cross-lingual…
- Users need only 3 seconds of audio for zero-shot timbre cloning,with cloned timbre similarity to…
- The model supports 14 languages including Chinese and English and is the first to enable emotion…
- It is built on a GPT-style semantic large model,SSL pretrained features with an ECAPA-TDNN speaker…
Quick Facts
- Published
- Updated
- Source
- IT之家·人工智能
- Category
- Model Releases
- AI score
- 72
网易有道把语音克隆的门槛压到了 3 秒,跨 14 种语言还能保持无口音,而且全量开源、商用无限制,对多语种配音和短剧出海是直接可用的工具。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: NetEase Youdao releases Confucius4-TTS: open-source 14-language cross-lingual accent-free voice cloning model. June 23, 2026. https://new.nat6.net/en/news/cmqqjmnoj0af8slp5kb7ky21w/