search

drethage / speech-denoising-wavenet

A neural network for end-to-end speech denoising
MIT License
673 stars 165 forks source link

got wrong array shape #6

Closed tonyyyye closed 6 years ago

tonyyyye commented 6 years ago

we have trained using our test data, but there are errors reported as

ValueError: Error when checking : expected condition_input to have shape (None, 1) but got array with shape (1, 5)

I wonder what the errors may be like, how we can fix it, and how are the designated target wav can be created. Noted that we have only clean wav files with 16000 sample rate.

Thanks

tonyyyye commented 6 years ago

i think i have fixed that by ffmpeg to resample it in the same format, but still i am not sure if this data preparation method is suitable for training. At least my training result is not as good as the demo one. Hope you can help me about that. Thank you.

tonyyyye commented 6 years ago

we have also encountered another problem showing that the result is blank with SNR 0.00000dB.

drethage commented 6 years ago

Hi tonyyye,

the ValueError is thrown because the model expects a scalar condition per training sample, you are trying to supply it with a 1x5 vector.

You don't need to preprocess the training data. The wav loading function will downsample audio to 16kHz automatically.