Microsoft releases VibeVoice-ASR-Streaming
AI summary
Microsoft released VibeVoice-ASR-Streaming, a unified streaming automatic speech recognition model that transcribes speech as audio arrives. According to Microsoft’s VibeVoice GitHub repository, it identifies who said what, supports customized hotwords and 10 languages, and emits text for each audio chunk.
In-depth insight
The release adds a streaming ASR model that combines transcription with speaker attribution and customizable terms, while Microsoft provides model links, documentation and a technical report.
Open-Source Frontier Voice AI

https://microsoft.github.io/VibeVoice/
🎙️ VibeVoice: Open-Source Frontier Voice AI
