Jiaqi Li

Hi, thanks for your suggeston. We're closing this issue for the reasons mentioned by @lmxue and @VocodexElysium , where we tried to use the default vocoder but resulted in poor...

Hi, we're closing this thread since the question about number of speakers in FACodec training is not directly related to this repo, for more detailed training we recommend looking at...

Hi @TechInterMezzo , we're releasing the FACodec training code in Amphion (it's now in a PR #229 and will be merged soon). You're welcome to check it out. Thanks!

We are actively developing a faster version together with more fast tts models, thanks for the attention and we'll release them soon before new year

> OMG! The time is 1200S when I ran the maskgct demo. > > Could you kindly check that you're using gpu for inference, otherwise it'll be very slow. Thanks!

@codename0og Thanks for your patience! It seems the author @VocodexElysium hasn't responded yet, some suggestions from me are: 1. https://github.com/NVIDIA/BigVGAN configurations since they have 44k config 2. check the long...

Thanks @CriDora for your feedback. I think your solution is correct, we'll try to debug this issue. Thanks!

Thanks for letting us know the issue. We have located the problem, to fix it you could set line 437 in https://github.com/open-mmlab/Amphion/blob/main/models/tts/valle_v2/base_trainer.py from "epoch_sum_loss += loss" to "epoch_sum_loss += loss.item()"....

Hi@CriDora, thanks for using our code! From my experience, the nccl timeout problem is usually related to a problem in the dataset loading, like it takes too long time to...

Please kindly refer to the audiolm and soundstorm paper for their implementation, which I understand is more than a single model. Thanks!