korean, depression/normal audio data set

[alexxony]

I have korean audio data sets which are labeled as depression and normal.

And each of them are at least 2 minutes.

Can I apply this model??

[frankenjoe]

As a start I would suggest you extract embeddings with the model and use them as features to train some classifier, e.g. a SVM. This should give you an idea if the model is applicable to your problem. In a next step you could try to fine-tune the model on your data.

[alexxony]

As a start I would suggest you extract embeddings with the model and use them as features to train some classifier, e.g. a SVM. This should give you an idea if the model is applicable to your problem. In a next step you could try to fine-tune the model on your data.

and How can i use gpu?? it took too long time

[frankenjoe]

https://audeering.github.io/audonnx/usage.html#run-on-the-gpu