tianjiao du

the test auido is 32-channel 2**15-length, for the batch 2 Besides, the num of trainable paras of the text condition generationis only 672M when follow the paper setting(text embding dim...

In hugging faces app and code, the durantion of the the generated samples must `% 2.5 == 0`，could someone explain it？

After just using VAE reconstruct a audio, I only get noise

5

Here is my code. Is there something wrong on my method about using vae? ``` `def recon_vae(self, filename): """ recon audio only by vae """ with torch.no_grad(): ``` waveform, sample_rate...