search

maxrmorrison / torchcrepe

Pytorch implementation of the CREPE pitch tracker
MIT License
407 stars 63 forks source link

.predict does not work. #10

Closed kkp15 closed 3 years ago

kkp15 commented 3 years ago

Hello! Thank you for your contribution :)

I was going to simply test the torchcrepe.predict() but it won't work. The code is simple as below. I installed the torchcrepe using pip.

audio, sr = torchcrepe.load.audio(path)

frame_rate = num_frames / (len(audio)/sr) crepe_step_size = 1000 / frame_rate # milliseconds fmin = 50 fmax = 550

model = 'full'

device = 'cuda:0'

batch_size = 1

pitch, harmonicity = torchcrepe.predict(audio, sr, crepe_step_size, fmin, fmax, model, batch_size=batch_size, device=device, decoder=torchcrepe.decode.viterbi, return_harmonicity=True)

TypeError: predict() got an unexpected keyword argument 'device'

Is this error expected to happen?

maxrmorrison commented 3 years ago

Thanks. The GitHub repo was a few commits ahead of PyPi. The device parameter was added in fd8325 to allow for processing audio signals larger than GPU memory. I just synced PyPi with the current commit, so reinstall, make sure you have version 0.0.12, and it should work. The README corresponding to the current PyPi version can always be found at https://pypi.org/project/torchcrepe/.