SReC

SReC copied to clipboard

@caoscott @chaoyuaw @philkr @HappyFacade I need your opinion: This should theoretically work to be trained on mel spectrograms and compress it and thus able to be used in audio compression...

trholding

root@27e30a88143c:/SReC# python3 -um src.encode Using torchac: True ['/SReC/src/encode.py'] Loaded model from models/openimages.pth. Epoch: 49 ModuleList( (0): ModuleList( (0): Identity() (1): Upsampler( (0): Conv2d(64, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))...

watertianyi