Error when loading the gru embedder

[Jayson Ng]

import torch
model = torch.hub.load('RF5/simple-speaker-embedding', 'gru_embedder')
model.eval()

When I run the above lines, I got the following error:

Using cache found in /root/.cache/torch/hub/RF5_simple-speaker-embedding_master
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
[<ipython-input-3-f8b105db465a>](https://localhost:8080/#) in <cell line: 4>()
      2 
      3 import torch
----> 4 model = torch.hub.load('RF5/simple-speaker-embedding', 'gru_embedder')
      5 model.eval()

6 frames
[~/.cache/torch/hub/RF5_simple-speaker-embedding_master/stft.py](https://localhost:8080/#) in __init__(self, filter_length, hop_length, win_length, window)
     33             # get window and zero center pad it to filter_length
     34             fft_window = get_window(window, win_length, fftbins=True)
---> 35             fft_window = pad_center(fft_window, filter_length)
     36             fft_window = torch.from_numpy(fft_window).float()
     37 

TypeError: pad_center() takes 1 positional argument but 2 were given

How could I resolve this bug?

[Jayson Ng]

Solved it by

pip install librosa==0.9.2

I think dependencies of the model should be freezed in a requirements.txt

[Matthew Baas]

Will do, thanks for the info.