They worked for me when training tacotron iirc

The current PyTorch install for TalkNET supports CUDA capabilities sm_37 sm_50 sm_60 sm_70

Installing the latest version might fix the issue

Can't say for sure tho cuz I haven't tested it

Ohh

Hey there. I was trying to train a new voice with TalkNet and was getting an error. To see where the problem was I decided to try and retrain a voice which I had already successfully trained. I got the same error. This is the error I'm now getting on a dataset I already trained successfully and it's the same error I get when I try to train a new voice. 100% 52/52 [00:13<00:00, 4.56it/s] 0 9/9 [==============================] - 12s 28ms/step --------------------------------------------------------------------------- AttributeError Traceback (most recent call last) in 145 print(str(i)) 146 audio_path = json.loads(l)["audio_filepath"] --> 147 f0_data[Path(audio_path).stem] = crepe_f0(audio_path) 148 149 # calculate f0 stats (mean & std) only for train set 2 frames /usr/local/lib/python3.7/dist-packages/crepe/core.py in to_viterbi_cents(salience) 142 143 # fix the model parameters because we are not optimizing the model --> 144 model = hmm.CategoricalHMM(360, starting, transition) 145 model.startprob_, model.transmat_, model.emissionprob_ = \ 146 starting, transition, emission AttributeError: module 'hmmlearn.hmm' has no attribute 'CategoricalHMM'

Hello I'm receiving an error on step 6 in Talking Training, how do I fix this?

Ah never mind I figure it out

i kept getting iton 5

Hi I have the same error that occured few days ago. Talknet doesn't find the reference audio.

use this nb: https://colab.research.google.com/github/justinjohn0306/TalkNET-colab/blob/main/Controllable_TalkNet.ipynb

I have the same error. I think I was already using this one

You gave this link last time

I also have another error. Each time I want to run the interface, I have to manually uninstall and install numpy

weird

I'll check it out then

Thank you Justinjohn-03

keep getting this nameerror for “nemo” on step 6

Did you miss step 4?

I kept getting it on step 5

Also I’m not at home at the moment

I just ran some tests and found out that keeping your audio length under 20 seconds works for some reason after Google updated some internal files recently

The issue won't occur if you run it locally on your computer

Oh Justin also I kept getting a error on step 5 when I tried training Kurt cobain with 77 wavs and 5 minutes and it was saying wav one not found also I cannot show error I’m not home at the moment

again...works fine for me

Yeah

you must have missed something

Maybe

But I’m not home tho

Cause that error also happened to me while I tried to make a talknet version of puff puff from yfm