search

bshall / UniversalVocoding

A PyTorch implementation of "Robust Universal Neural Vocoding"
https://bshall.github.io/UniversalVocoding/
MIT License
237 stars 41 forks source link

path to waveform directory #2

Closed maximilianchen closed 5 years ago

maximilianchen commented 5 years ago

Hi @bshall, I wondered how to set the path to the downloaded waveform directory when preprocessing, as it not as the parameter.

bshall commented 5 years ago

Hi @maximilianchen, I wrote the preprocessing.py file specifically with the structure of the ZeroSpeech2019 dataset in mind. If you're looking to train on your own dataset the easiest approach would be to change the line here to point to your directory. For example, if your .wav files are in the directory ~/dataset/wavs/ you should be able to change the line to

wav_dirs = ["~/dataset/wavs/"]

I hope that helps.

I can also add a command line argument if that makes things simpler?

maximilianchen commented 5 years ago

Hi @maximilianchen, I wrote the preprocessing.py file specifically with the structure of the ZeroSpeech2019 dataset in mind. If you're looking to train on your own dataset the easiest approach would be to change the line here to point to your directory. For example, if your .wav files are in the directory ~/dataset/wavs/ you should be able to change the line to

wav_dirs = ["~/dataset/wavs/"]

I hope that helps.

I can also add a command line argument if that makes things simpler?

Thanks a lot for your information, @bshall . I had missed out the line you mentioned above when I brought up the issue. Now I have managed to point to the wav path.

bshall commented 5 years ago

No problem. Let me know how the training goes. I haven't tested the model out on other datasets so I'd be interested to hear how it goes.