This model requires you to accept the MIT license terms before download.
Description
VibeVoice 1.5B — Microsoft text-to-speech model for long-form, multi-speaker audio such as conversational podcasts, producing expressive voices with natural turn-taking between speakers. MIT license.
Tags
At a glance
| Kind | text-to-speech model |
|---|---|
| Precision | bfloat16 |
| Container size | 5.0 GiB |
| Weights inside | 5.0 GiB |
| Components | 6 |
| Licence | MIT |
| Origin | microsoft/VibeVoice-1.5B @ c00898d257e6 |
What is inside the container
A .nbx container packs every part of VibeVoice-1.5B — 6 components with weights, graphs and runtime settings — so one file is all the runtime needs.
| Component | Blocks | Layers | Weights |
|---|---|---|---|
| model.acoustic_connector | 0 | — | 4.7 MiB |
| model.acoustic_tokenizer | 0 | — | 1.3 GiB |
| model.language_model | 0 | 28 | 2.9 GiB |
| model.prediction_head | 0 | — | 235 MiB |
| model.semantic_connector | 0 | — | 4.9 MiB |
| model.semantic_tokenizer | 0 | — | 657 MiB |
Default settings
The values the container ships with. They apply when you do not pass your own.
- Sample rate (Hz)
- 24000
- Temperature
- 0.7
Memory needed
The weights of VibeVoice-1.5B take 5.0 GiB in bfloat16. That is the floor for the memory the runtime has to place, spread over the devices it finds; activations come on top and grow with the size and length you ask for. A measured peak per device class is not published for this model yet.
Verify what you downloaded
Every download carries its SHA-256, and the file endpoint sends it in the X-NBX-SHA256 header. The container is 5.0 GiB (5,412,514,386 bytes).
$ sha256sum VibeVoice-1.5B.nbx
173b37ca46d4569c51099fdb85114b342d5c990d969b3ba8ae72086229801543
Questions
What is VibeVoice-1.5B?
A text-to-speech model by microsoft, distributed as a bfloat16 .nbx container and licensed MIT.
How big is the download?
5.0 GiB, of which 5.0 GiB are weights.
How do I run it?
Install the runtime from the download page, then use the commands under “CLI Usage” below: neurobrix import microsoft/VibeVoice-1.5B, then neurobrix serve --model VibeVoice-1.5B.
CLI Usage
$ neurobrix import microsoft/VibeVoice-1.5B
$ neurobrix serve --model VibeVoice-1.5B
$ neurobrix run --prompt "your prompt here"
$ neurobrix stop