Unverified80% confidenceFactTime unknown
Google的语音识别引擎支持超过125种语言，底层采用端到端的Transformer架构模型
1
Sources
80%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
Unverified该项目技术栈选择微调多语言模型作为骨干，使用Whisper做语音识别和NLLB做翻译65% similarVerifiedIn 2022, the Google Brain team first systematically documented the phenomenon of 'emergent abilities' in large language models in the paper 'Emergent Abilities of Large Language Models'64% similarUnverifiedGoogle 在 Translatotron 系列论文中探索了端到端语音翻译，证明其在保留说话人声纹、语速和情感信息方面具有优势并降低延迟64% similarUnverified2006年Google翻译初始系统使用了一个在数万亿词文本上训练的5-gram语言模型，这很可能是世界上第一次万亿token规模的语言模型训练63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/23148API
curl https://kongchang.com/api/v1/knowledge/claims/23148MCP
get_claim(id=23148)