# tacotron-2-support
- j
Justin
10/29/2022, 4:35 PMTurn it on - j
Justin
10/29/2022, 4:36 PMSee if that helps - c
CmanflipYT
10/29/2022, 4:36 PMOh alright. - c
CmanflipYT
10/29/2022, 4:36 PMI'll try that next time. I'm currently training Madame Blueberry in talknet form - a
AhmadGT
10/29/2022, 4:36 PMany ideas how to fix? - j
Justin
10/29/2022, 4:37 PMYou can use the CPU synthesis nb - c
CmanflipYT
10/29/2022, 4:38 PMOk - a
AhmadGT
10/29/2022, 10:31 PMis when the Attention/val is orange straight line/diagonal i know that is the model is traind well? - r
Radak
10/30/2022, 12:08 AMI can’t submit the model it keeps doing this can someone fix it - i
I love the hive
10/30/2022, 12:34 PMPls help i using 100 wavs and try to generate mel spectrogram it give me error:list index out of range - c
Cris140
10/30/2022, 3:56 PMblank line at the txt file - j
Justin
10/30/2022, 4:15 PMalso, wav number 3 has no punctuation - a
AhmadGT
10/30/2022, 7:13 PM؟ - i
I love the hive
10/31/2022, 4:49 AMthanks i decided to check and yes there were blank lines. - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 8:09 PMIf anyone can; I still need help with this project please. I haven't touched the AI project since this post; and I'm worried when I pick it up again I'd still get immediate errors when trying to run the training model. - r
Reclezon
11/01/2022, 9:01 PMIdk if I'm actually looking at the right post (Discord is dumb) but just on what I can find from chat looks like the legacy trainer? - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 9:02 PMYes, I am using the legacy trainer - r
Reclezon
11/01/2022, 9:07 PMI would suggest the pipeline one as that one should be a better experience. For legacy though, setup cells gave no errors, right? - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 9:26 PMThe short of it is I would run all the required steps needed; no errors would pop up; I pressed the last button to begin training the model, And it instantly wouldn't let me May I have the link to the Pipeline model please? - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 9:27 PMHeres the error I kept getting - r
Reclezon
11/01/2022, 9:32 PMI've been meaning to look into pickle but unfortunately Im not able to help, and Idk python much either. The pipeline nb is in #841437191073955920 , other wise here's a faster link. https://colab.research.google.com/drive/1jF-Otw2_ssEcus4ISaIZu3QDmtifUvyY - r
Reclezon
11/01/2022, 9:34 PMFor uploading the dataset to pipline, you can use a link, or upload it. Zip should be formatted asCopy codedataset/Transcription.txt dataset/wavs/[x amount of wavs] - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 10:05 PMThanks <:) I'll get into it in just a minute - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 11:03 PMi won't lie, I'm tad confused on where to do this, I'm on step 2 rn, and this is what my zip looks like - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 11:04 PMI am a big smooth brain, complete dumb dumb on programing, feel free to poke fun if you like - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 11:07 PMthe short of it is that I used to just be able to load the audio and transcripts into the side folder, not quite linking different google drive links together and,,yeah - r
Reclezon
11/01/2022, 11:10 PMThat was me too - r
Reclezon
11/01/2022, 11:13 PMLoading the dataset manually is a bit different.. once you run the setup cells, open the file explorer on the sidebar, there should be a folder called
, that's where you'll upload the zip, which should look similar to the example attached. I haven't done it through a link, but from what I've heard, it should he similar.UPLOAD_HERE - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 11:14 PMThank you Soo much!🫂 - u
🏝✍Berkeley•Ray•Brewski
11/01/2022, 11:42 PMtrying to get the dataset rn