George Grigorev

thank you, `pip install pytorch-lightning==1.0.8 omegaconf==2.0.0` helped for some reason `conda env create -f environment.yaml` didn't install some required libs

have you tried to train on LJSpeech or your dataset? How much iterations needed comparing with HiFiGAN? Do you have checkpoints somewhere?

got it, thanks

tried it out. i compare publicly available universal v1 hifigan (trained on 2.5M iterations on vctk) with this one trained on 150k at new HIFI-TTS dataset (5 times more data)....

3x3090 with batch 16 but I can confirm that fre-gan is training much faster than hifi-gan

@EmreOzkose i guess that's because author shared discriminator only for universal model

@balag59 same here. But I generated mels from tts model. After 50k iterations overall quality of audio increased, but identity of speaker lost. I guess I should train a lot...

 @jik876 I guess it's not normal and I should grab more data?

I had the same issue when SD2 only came out. I believe it has something with v_prediction of 768 model. I tried to modify sampler in dreambooth training script but...

alright. before: https://voca.ro/1meOfwM2dIEw after:https://voca.ro/11L19CihKeI6