ValueError n_components=128 must be between 0 and min(n_samples, n_features)=2 with svd_solver='full'

[lang216]

Hi,

I was using the google Colab provided and tried to train on some sounds. However, after running the training block, I always get the error ValueError n_components=128 must be between 0 and min(n_samples, n_features)=32 with svd_solver='full'.

I think it has something to do with the PCA algorithm, more specifically, the number of samples I used. Do I have to use more 128 samples in order to make this model work?

Best,
Lang Chen

[ghost]

i had similiar issue and i think it was because my training dataset was too small / not enough audio material #108

[lang216]

Yes, this is the issue! We need to use much more samples (larger training set)!

[clusternd]

Have you figured out approximate minimum size of training dataset? I've just run into the same problem

[lang216]

I think you need more than 128 samples. Also, you might want your total audio duration to be more than 3 hours in order to have a good result.