Resources & links
Official source, weights, a live demo and further reading. Filter by type or search.
| Resource | Type | Description | Link |
|---|---|---|---|
| GitHub source | Official | microsoft/VibeVoice — code, docs and fine-tuning resources. | https://github.com/microsoft/VibeVoice |
| Hugging Face (microsoft) | Weights | All six open checkpoints and their model cards, under the Microsoft org. | https://huggingface.co/microsoft |
| ASR Playground | Demo | Try VibeVoice-ASR in the browser. | https://aka.ms/vibevoice-asr |
| Project page | Official | Demos, samples and an overview of the whole family. | https://microsoft.github.io/VibeVoice |
| ASR-Streaming docs | Official | How to run the streaming build, including the FastAPI/WebSocket demo. | https://github.com/microsoft/VibeVoice/blob/main/docs/vibevoice-asr-streaming.md |
| VibeASR.cpp | Official | The edge CPU inference engine behind the BitNet build. | https://github.com/microsoft/VibeASR.cpp |
| Streaming ASR demo | Demo | Try live, speaker-attributed transcription in the browser. | https://aka.ms/vibeasr |
| Azure AI Foundry Labs | Demo | Explore VibeVoice-ASR through Microsoft Foundry, no local setup. | https://labs.ai.azure.com/innovations/vibevoice-asr/ |
| TTS paper (arXiv 2508.19205) | Paper | Expressive podcast generation with next-token diffusion. ICLR 2026 Oral. | https://arxiv.org/abs/2508.19205 |
| ASR report (arXiv 2601.18184) | Paper | The 60-minute single-pass recognition model. | https://arxiv.org/abs/2601.18184 |
| ASR-BitNet report (arXiv 2607.21075) | Paper | Heterogeneous quantization for CPU inference: 4.62 GB to 1.58 GB. | https://arxiv.org/abs/2607.21075 |
| ASR-Streaming report (arXiv 2609.02812) | Paper | LLM-based end-to-end streaming speaker-attributed recognition. | https://arxiv.org/abs/2609.02812 |
| Reference article (itnotetk) | Reading | Chinese write-up on the multi-speaker long-form TTS. | https://www.itnotetk.com/2026/05/01/vibevoice-microsoft-multi-speaker-tts/ |
| AllAboutAI review | Reviews | Hands-on review with a user-survey score (4/5). | https://www.allaboutai.com/ai-reviews/microsoft-vibevoice/ |
| Applied AI Tools — user review analysis | Reviews | Open-source explainer plus aggregated user reviews. | https://appliedai.tools/ai-for-content/microsoft-vibevoice-tts-open-source-explained-with-user-review-analysis/ |
| Hacker News discussion | Reviews | Community thread with praise and critical takes. | https://news.ycombinator.com/item?id=45114245 |
| Slator coverage | Reviews | Language-industry coverage of the long-form model. | https://slator.com/microsoft-research-vibevoice-long-form-speech-synthesis/ |
| OpenAI — next-gen audio models | Compared | gpt-4o-transcribe and gpt-4o-mini-tts announcement. | https://openai.com/index/introducing-our-next-generation-audio-models/ |
| OpenAI — gpt-realtime | Compared | OpenAI's production realtime speech-to-speech models. | https://openai.com/index/introducing-gpt-realtime/ |