Haohe (Leo) Liu / 刘濠赫 comments

Results 20 comments of


Haohe (Leo) Liu / 刘濠赫

The validation loss seems too high

The same phenomenon in your log file: https://github.com/YuanGongND/ast/blob/master/egs/audioset/exp/test-full-f10-t10-pTrue-b12-lr1e-5/log_2090852.txt

The validation loss seems too high

Many thanks for the explanation! Haohe

nn split

@shenbuguanni The subband analysis and synthesis layer are realized using the torch.nn.module and 1-d convolution. So you can set them as trainable or frozen, both will work.

Doubt on how to use it in practice

Hi @lucasbr15 , Torchsubband is a tool that computes a new representation of the audio. It's definitely usable in other architecture as long as you design proper architecture. I build...

Pre-trained Models Download problems

@YongyuG Thanks for your suggestions. The download link on baidu netdisk is updated in the readme. Hope you enjoy this repo. Haohe

"voicefixer" is not recognized as an internal or external command

@Scudoxx Sorry for the inconvenience. Now the bug on the windows machine should be fixed. Please refer to the [Command line](https://github.com/haoheliu/voicefixer#command-line) section of the README. Hope you enjoy it. And...

FileNotFoundError: [Errno 2] No such file or directory: 'C:\\Users\\sadness112\\.cache/voicefixer/analysis_module/checkpoints/vf.ckpt'

Hi @sadness112, sorry for the error. It seems the error happens on loading model weight. The system cannot identify the proper path. I guess this is because of the Windows...

FileNotFoundError: [Errno 2] No such file or directory: 'C:\\Users\\sadness112\\.cache/voicefixer/analysis_module/checkpoints/vf.ckpt'

@justinjohn0306 Thanks for pointing it out!

Streamlit says Numpy isn't available when I already have it installed.

Sorry for the late reply. This seems to be a problem in the running environment. Have you solved that issue now?

RuntimeError: PytorchStreamReader failed reading zip archive: failed finding central directory

Sorry for the late reply. It's hard to tell the reason with the information you provided. I haven't tested VoiceFixer on Windows so that might be a potential issue. I...