PulseAugur
EN
LIVE 14:52:02
中文(ZH) 网易有道首发14语种零口音语音克隆模型,无需参考文本即可复刻任意音色

NetEase Youdao releases open-source 14-language voice cloning TTS model

NetEase Youdao has launched Confucius4-TTS, a new large model TTS engine that supports 14 languages. This engine is notable for its ability to clone voices with zero-shot learning, requiring only 3 seconds of audio and no reference text to replicate a speaker's tone and emotion. The model is fully open-source, with weights and tools available for local deployment, aiming to reduce costs and barriers for creators and developers in areas like digital humans and cross-lingual communication. AI

IMPACT Enables low-cost, high-quality voice cloning and cross-lingual synthesis, potentially accelerating adoption in digital content creation and global communication.

RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on 雷峰网 (Leiphone) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

NetEase Youdao releases open-source 14-language voice cloning TTS model

COVERAGE [1]

  1. 雷峰网 (Leiphone) TIER_1 中文(ZH) ·

    NetEase Youdao first releases 14-language zero-accent voice cloning model, able to replicate any voice without reference text

    <p>当前,人工智能作为培育新质生产力的核心引擎,已上升为国家战略层面。国务院《关于深入实施“人工智能+”行动的意见》明确提出,要加快AI核心技术自主创新、降低产业落地门槛、构建开放共享的国产AI生态,推动人工智能与千行百业深度融合。</p><p>在这一战略背景下,网易有道正式推出“子曰4.0”大模型体系TTS语音合成引擎——Confucius4-TTS,并已面向全球用户开放。近日,该引擎凭借全球首个不依赖参考文本即可实现14语种无口音跨语种语音克隆的开创性突破引发行业高度关注,为数字人、跨境传播、智能教育等产业提供国产化、低成本语音克隆功能。</p><…