# tacotron-2-support
- a
AppData 174
08/19/2022, 2:10 PMI'm retraining and the graph looks much better - u
(Dawn) Will Draw Fictional Women
08/19/2022, 2:10 PMbecause thats whats in your transcript - m
milesmorales
08/19/2022, 2:11 PMwould this fix the issue - u
(Dawn) Will Draw Fictional Women
08/19/2022, 2:11 PMidk - u
(Dawn) Will Draw Fictional Women
08/19/2022, 2:11 PMprobably not - m
milesmorales
08/19/2022, 2:16 PMwhats the issue - m
milesmorales
08/19/2022, 2:16 PMughhhhhhhhhhh - u
(Dawn) Will Draw Fictional Women
08/19/2022, 2:17 PMyour transcript - u
(Dawn) Will Draw Fictional Women
08/19/2022, 2:17 PMi just told you - m
milesmorales
08/19/2022, 2:17 PMwhat about it - m
milesmorales
08/19/2022, 2:18 PMthe name of the transcript is just list - m
milesmorales
08/19/2022, 2:18 PMdo i call it filelist then - m
milesmorales
08/19/2022, 2:23 PMnvm i figured it out - m
milesmorales
08/19/2022, 2:26 PMoh god how do i fix this - m
milesmorales
08/19/2022, 2:26 PMrename all files to just wave - l
Linkle Ze
08/19/2022, 2:44 PMSo I was wondering if you need the tensor board to make a model? I keep getting no dashboards are active for the current data set and when I move on, a new model isn't created after I press train. - l
Linkle Ze
08/19/2022, 2:47 PMI also get FileNotFoundError: [Errno 2] No such file or directory: 'speakers/voice/270.wav.wav' - u
(Dawn) Will Draw Fictional Women
08/19/2022, 2:49 PMtranscript error - l
Linkle Ze
08/19/2022, 2:49 PMYeah, I just checked it and for some reason .wav was repeated twice. Fixed that so we'll see how it goes now - l
Linkle Ze
08/19/2022, 2:54 PMRuntimeError: shape '[1, 1, 106126]' is invalid for input of size 212252 - l
Linkle Ze
08/19/2022, 2:54 PMWhat does this error mean? - h
haru0l
08/19/2022, 5:29 PM(some?) of your audio is in stereo - h
haru0l
08/19/2022, 5:29 PMyou can run the autofix thing the pipeline notebook has - l
Linkle Ze
08/19/2022, 8:37 PMAh, thank you. - r
Radak
08/20/2022, 2:53 AMI’m trying to make a cover to Billionare with the xxxtentacion ai it’s having some error the audio is 15 seconds - r
Radak
08/20/2022, 2:56 AMI figure it out I actually had the audio and I got it fixed never mind - a
AppData 174
08/20/2022, 5:31 AMIf he length is exactly 15 seconds then it will not work, you need to trim it a little bit Also, it's a talknet model, not a tacotron (look at the name of this channel) - r
Radak
08/20/2022, 6:05 AMI know its a talknet model - c
Cris140
08/20/2022, 10:46 PMIs it better to ignore this layers or just leave "null"? ["speaker_embedding.weight", "spkr_lin.weight", "spkr_lin.bias"], - u
{K EY1} (Kei)
08/20/2022, 11:04 PMIf you're warm starting, ignore the layers If you're resuming from a checkpoint, put to null