search

slp-rl / aero

This repo contains the official PyTorch implementation of "Audio Super Resolution in the Spectral Domain" (ICASSP 2023)
MIT License
190 stars 24 forks source link

AssertionError in training #18

Closed yxlu-0102 closed 7 months ago

yxlu-0102 commented 8 months ago

Hello, I ran into this problem when training the aero model: assert len(self.hr_set) == len(self.lr_set) AssertionError

yxlu-0102 commented 8 months ago

I should be caused by segmenting, can I set segment=None?

m-mandel commented 8 months ago

It's been a while since I looked at the code, but if I recall correctly, setting segment=None will cause the dataset to get each sample as a whole. You are welcome to dive into the code here: https://github.com/slp-rl/aero/blob/63fc8a26d96ea0ce39d3a17ef535f4eff186e9ff/src/data/datasets.py#L130.

If you need more assistance, please share the config files you used, and print out the actual length of the hr_set and lr_set you are receiving.

yxlu-0102 commented 8 months ago

Have you ever trained the model on the condition of 2-16? If you have, can you provide the pre-trained models?