Unverified50% confidenceFactExact time
Clef 属于音频-语言模型（Audio-Language Model）这一细分方向
1
Sources
50%
Confidence
Long-term
Relevance
10/4/2026
First Seen
Sources
Related Entities
Related Claims
Unverified微软VALL-E将语音合成重新定义为语言建模问题，把语音编码为离散声学标记并像大语言模型预测下一个词一样预测下一个声学标记68% similarUnverifiedCline本身不含AI模型，需要接入一个大语言模型作为后端67% similarUnverified跨语言词义消歧（CLWSD）不同于单语词义消歧，需要在源语言多义词与目标语言多个可能翻译之间做选择65% similarUnverified端到端语音翻译模型可部分保留说话人的韵律特征等副语言信息65% similarUnverifiedLAPE的核心理念是以大语言模型提取的语言表征为锚点，再将副语言线索融入其中64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/971694API
curl https://kongchang.com/api/v1/knowledge/claims/971694MCP
get_claim(id=971694)