audeering/w2v2-how-to

korean, depression/normal audio data set

alexxony opened this issue · 3 comments

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??

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.

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