How to extract voice embeddings? about espnet

[ftshijt]

Hi @maxpain, thanks for your following up. We have not yet officially released the colab. The colab I shared is mostly for tentative view (though the API to embeddings is fixed already, so feel free to refer to the usage examples in the notebook).

We expect to release the official notebook with regarding to our ESPnet-SPK paper in around two month later.

[sw005320]

Thanks for your questions.
@Jungjee, can you answer it?
We have a nice Python interface to extract features, but we may need to refine our documents.

[ftshijt]

@maxpain We have a colab demonstration for the embedding extraction, you may check this https://colab.research.google.com/drive/1gizVM0T0O41cgY_sEHkf7eN-Xxl5YbWB?usp=sharing

[maxpain]

https://colab.research.google.com/drive/1gizVM0T0O41cgY_sEHkf7eN-Xxl5YbWB?usp=sharing

Thanks!
How exactly this colab should be found in the docs? I didn't find it, I only found this repository, but there are only assignments.

[maxpain]

Thanks!

[maxpain]

@sw005320 @ftshijt do you have some colab for speaker Diarization tasks?