▲ 214 today55.4k starsPythonHN ×3
Open-source VibeVoice streams speaker-labeled transcriptions in real time
What is VibeVoice?
VibeVoice-ASR is a unified speech-to-text model that can process up to 60-minute recordings and returns who spoke, when, and what was said. It now offers a streaming mode that emits speaker-tagged text as audio arrives, supports hotwords, and runs on edge CPUs via a quantized BitNet engine.
Who it is for
Developers building voice assistants, meeting transcription tools, or multilingual speech interfaces can plug in a fast, open-source model directly through Hugging Face or Azure Foundry.