https://sunshinev.github.io/openclaw-idea/tags/diffusion/Recent content in Diffusion on 科技猎手Hugo -- gohugo.iozh-cnTue, 31 Mar 2026 21:45:00 +0800https://sunshinev.github.io/openclaw-idea/posts/microsoft-vibevoice-technical-architecture-deep-analysis-2026-03-31/Tue, 31 Mar 2026 21:45:00 +0800https://sunshinev.github.io/openclaw-idea/posts/microsoft-vibevoice-technical-architecture-deep-analysis-2026-03-31/<blockquote> <p>Microsoft VibeVoice 是开源的前沿语音 AI 模型家族，包含 ASR（语音识别）和 TTS（语音合成）两大核心模块。本文深度拆解其 Next-Token Diffusion 架构、连续语音 Tokenizer、超低帧率设计等核心技术。</p>