jishengpeng / WavTokenizer

SOTA discrete acoustic codec models with 40 tokens per second for audio language modeling
MIT License
787 stars 43 forks source link

Files Missing? #39

Open goforher opened 1 month ago

goforher commented 1 month ago

In line 15 of ac.py, the code from ..binary import BitPacker, BitUnpacker references binary, but there is no such folder or dependency package in the project.

jishengpeng commented 1 month ago

In line 15 of ac.py, the code from ..binary import BitPacker, BitUnpacker references binary, but there is no such folder or dependency package in the project.

This may not seem important

goforher commented 1 month ago

OK, thanks

goforher commented 1 month ago
image

Hello, I have a question. Why is the forward method in this class not using the lines of code # residual = residual - quantized.detach() and # quantized_out = quantized_out + quantized, while the encode and decode methods do? Won't this cause inconsistency between training and inference?

jishengpeng commented 1 month ago

image Hello, I have a question. Why is the forward method in this class not using the lines of code # residual = residual - quantized.detach() and # quantized_out = quantized_out + quantized, while the encode and decode methods do? Won't this cause inconsistency between training and inference?

The two functions are not utilized; instead, the forward function is employed.

goforher commented 1 month ago

Alright, thank you.

goforher commented 1 month ago

Hello, when I try to retrain the model using the AISHELL-3 dataset, I encounter this error as shown in the picture. The configuration file is as follows; could it be due to an incorrect configuration file or is it caused by the dataset?

image image image image