how to adapt this model to separate different speakers

Thanks for sharing the amazing work!

How to adapt this model to separate different speakers？ e.g., five people in one audio. They are talking one-by-one instead of speaking together.

Start with the repository's inference code and example notebooks, then check how the model handles a single audio input containing multiple speakers. Done would require a documented, reproducible path for separating five sequential speakers, or a clear explanation of whether the current model supports that use case.

Written by the indexing model from the issue text.