Visit website ↗
Use ASR checkpoints for transcripts, timestamps and speaker segments, or a separate Realtime model for streaming English single-speaker TTS.
TOOLORA / Local reading edition
VibeVoice
Use ASR checkpoints for transcripts, timestamps and speaker segments, or a separate Realtime model for streaming English single-speaker TTS.
Overview
VibeVoice contains task-specific speech models. ASR takes long recordings and hotword context to jointly return text, timestamps and speaker segments; ASR-Streaming emits text as audio chunks arrive. Realtime-0.5B is a separate English single-speaker TTS model using prepared voice prompts, not the same capability set as long-form multi-speaker models. Official inference and serving examples are available. Reviewed ASR and Realtime cards declare MIT; scope remains model-specific.
- Company country / region
- United States · Published company address
Access options
Official sources and references
- Official websitehttps://github.com/microsoft/VibeVoice
- Documentationhttps://github.com/microsoft/VibeVoice/blob/main/docs/vibevoice-asr.md
- Documentationhttps://github.com/microsoft/VibeVoice/blob/main/docs/vibevoice-asr-streaming.md
- Documentationhttps://github.com/microsoft/VibeVoice/blob/main/docs/vibevoice-realtime-0.5b.md
Descriptions summarize public information; consult official websites for features, pricing and regional availability.