it was unintentioal

anyway, red (not the bird) has to be apoved

red the lil imp

I have a model of Red who isn't the

@Gosmokeless28 5volt sovits model

I'm not trying to be rude, but do you realize that gosmokeless has a life and he can't just approve everything the second you upload it right? He's doing his job correctly and is approving as he sees fit and in his own time, not yours. You're lucky your models end up on uberduck at all.

Welp this seems a little bad That dip is worrying me and I dont know why

I honestly have no idea at all what it means but ive been told it's important

..

i am patient in the virtual workd, but not in the real world

This virtual world is still ran by real people though; so kindness and manners apply just as much as in the real world

Probably stupid to ask but can so vits models suffer from overfitting like legacy tacotron2 models?

I would say probably so but I'm not entirely sure. The most steps I've ever trained is around 300,000 steps and it wasn't giving any artifacts, but I'm sure at some point it'll overtrain just as almost every other AI does

Hi, I am stuck; nothing happens when i press Infer; using local version of 3.14 so vits svc fork on Mac M1

can somebody help me with this error

2023-05-22 16:26:36.481292: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations. To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags. 2023-05-22 16:26:37.625217: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT Traceback (most recent call last): File "/content/so-vits-svc/train.py", line 330, in main() File "/content/so-vits-svc/train.py", line 47, in main hps = utils.get_hparams() File "/content/so-vits-svc/utils.py", line 416, in get_hparams with open(config_save_path, "w") as f: PermissionError: [Errno 1] Operation not permitted: './logs/44k/config.json'

Why does it create a new pair of .pths at 266 epoches, instead of 130 epoches?

The sovits fork 4.0 notebook seems to be giving MAJOR issues with the pitch detection on the "Start Inference (For Advanced Users)" cell I can't figure out if this is something I'm doing wrong, or something up with the notebook, but the normal standard inference cell works perfectly.. https://cdn.discordapp.com/attachments/1097649007804821574/1111061306913599549/cantbuymelove_Cabby1_pitch0_dio_8.wav

For reference, that's the original

If I use my dataset used for training a talknet model, would it work for this too?

Yes

Except from what I've heard, you won't need the transcript file for this one?

Right

And the audio clips can also be in 48kHz

Oh cool!

what does a cluster model do? also, whats the reccomended amount of steps

go this error when trying to display inferred audio. the audio I'm trying to generate also isn't appearing anywhere. How can I fix this?