Mr Segfault

The error log is truncated here is the entire output start to finish: (Retrieval-based-Voice-Conversion-WebUI) user@machine:~/Retrieval-based-Voice-Conversion-WebUI$ python infer-web.py Use Language: en_US 16系显卡强制单精度 Running on local URL: http://0.0.0.0:7865 start preprocess ['trainset_preprocess_pipeline_print.py', '/home/user/Music/dataset_hj/40k/Minidata',...

Alright, volume wise I've run https://github.com/henrymaas/AudioSlicer/ on my dataset so any silences were already stripped out and then ran the samples into a gentle saturator plugin, they are quite punchy...

Another observation I had was that this exact dataset and computer with previous versions of RVC was able to train models (just not output an index file) but that training...

Still on my first PC with the much larger data set but it's saying a lot of samples are, for example: `1649_5.wav-contains nan` These samples had previously been recognized just...

I think this is now fixed as we're getting now further than this point but can leave it open for a day or two to let anyone else chime in...

Closing as this has been fixed. Thank you developers.

**Updated to the current version that does not need the total_fea.npy output and the error still persists:** INFO:train5:Saving model and optimizer state at epoch 20 to ./logs/train5/G_20.pth INFO:train5:Saving model and...

Tried again (4/27) with a current build, it looks like the Train Feature Index feature is still broken (not preparing index file) Logs are slightly different because the code has...

I was worried that it was my lower powered PC causing the issue so I tried it on my more powerful one, The logs are less lengthy but are still...

I was getting this feedback on a dual 3060 (12gb ea) as well as a 1660ti