error: The frames array is empty. No features to compute.

[linmi]

6_speakers.wav.zip

.cargo/registry/src/index.crates.io-6f17d22bba15001f/knf-rs-0.1.7/src/lib.rs:38:9

During my testing process, only 3 people will also encounter 7 8 9 and many ?. The original pyannote accurately identified it.

Can you provide an option that does not specify the number of people and let the system automatically determine it? thanks.

BTW, This is a great project, very fast.

[thewh1teagle]

.cargo/registry/src/index.crates.io-6f17d22bba15001f/knf-rs-0.1.7/src/lib.rs:38:9

I see the error. it happens because the specific segment is only 100ms. Maybe it's too short for the features extraction for embedding. Not sure yet how to handle this.

During my testing process, only 3 people will also encounter 7 8 9 and many ?. The original pyannote accurately identified it.

Can you share your snippet with pyannote so we can compare?

Can you provide an option that does not specify the number of people and let the system automatically determine it? thanks.

I updated the examples. Both now returns the speaker always if no error.

BTW, This is a great project, very fast.

Thanks!