can't i modify it just for me so i can train this night ?

it's late here too

yea i think i'll stick to the legacy tacotron2 for now

i just have to add the symbols, fix the cleaners and ignore the layers of the pretrained model

thanks for the quick replies

Mounted at /content/drive Copy the model path from the Files menu on the left and paste it here: https://drive.google.com/file/d/1fXzsXvqdUtnm31tRBIjdc1vqhMbudQDE/view?usp=sharing --------------------------------------------------------------------------- FileNotFoundError Traceback (most recent call last) in 28 most_recent_checkpoint_epoch = -1 ---> 29 for i in os.scandir(storage_path): 30 match = re.match(checkpoint_regex, i.name) FileNotFoundError: [Errno 2] No such file or directory: '/content/drive/MyDrive/tacotron' During handling of the above exception, another exception occurred: FileNotFoundError Traceback (most recent call last) in 40 drive.mount("/content/drive") 41 if not os.path.exists(most_recent_checkpoint_path): ---> 42 raise FileNotFoundError 43 44 from collections import OrderedDict FileNotFoundError: i got this 👆

I added a link to an archive with audio material that I uploaded to google disc and started learning, but in the end I got this error

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True). Copy the model path from the Files menu on the left and paste it here: /content/drive/MyDrive/Deathwing.RAR --------------------------------------------------------------------------- UnpicklingError Traceback (most recent call last) in 45 import torch 46 ---> 47 model = torch.load(most_recent_checkpoint_path, "cpu") 48 model_finalized = OrderedDict() 49 model_finalized["state_dict"] = model["model"] 1 frames /usr/local/lib/python3.9/dist-packages/torch/serialization.py in _legacy_load(f, map_location, pickle_module, **pickle_load_args) 1000 "functionality.") 1001 -> 1002 magic_number = pickle_module.load(f, **pickle_load_args) 1003 if magic_number != MAGIC_NUMBER: 1004 raise RuntimeError("Invalid magic number; corrupt file?") UnpicklingError: unpickling stack underflow

--------------------------------------------------------------------------------------------------------------------------------------------------------------- Now I got this error. Last time when I showed an error - I didn't correctly set the link to my archive, and only now, as the last noob, I understood how to make the correct link to the archive. But even in this case I got some error 😹

hello, when i train in pipeline tacotron2, this message appears, what could it be ?

It's not like that's an error.

ok

The regular tacotron2 needed like 500 wave files to train in order to sound decent. Does uberduck require as much?

if you're clever you can go down to morshu levels of audio

mostly because the pipeline notebook has better base models (the default is trained on a thousand voices so it's very generalized)

Really? That's very interesting.

ERROR: Unsupported Method : Neltharion/1.wav ERROR: Unsupported Method : Neltharion/10.wav ERROR: Unsupported Method : Neltharion/11.wav ERROR: Unsupported Method : Neltharion/12.wav ERROR: Unsupported Method : Neltharion/13.wav ERROR: Unsupported Method : Neltharion/14.wav ERROR: Unsupported Method : Neltharion/15.wav ERROR: Unsupported Method : Neltharion/16.wav ERROR: Unsupported Method : Neltharion/17.wav ERROR: Unsupported Method : Neltharion/18.wav ERROR: Unsupported Method : Neltharion/19.wav ERROR: Unsupported Method : Neltharion/2.wav ERROR: Unsupported Method : Neltharion/20.wav ERROR: Unsupported Method : Neltharion/21.wav ERROR: Unsupported Method : Neltharion/22.wav ERROR: Unsupported Method : Neltharion/23.wav ERROR: Unsupported Method : Neltharion/24.wav ERROR: Unsupported Method : Neltharion/25.wav ERROR: Unsupported Method : Neltharion/3.wav ERROR: Unsupported Method : Neltharion/4.wav ERROR: Unsupported Method : Neltharion/5.wav ERROR: Unsupported Method : Neltharion/6.wav ERROR: Unsupported Method : Neltharion/7.wav ERROR: Unsupported Method : Neltharion/8.wav ERROR: Unsupported Method : Neltharion/9.wav ERROR: Unsupported Method : Neltharion/list.txt --------------------------------------------------------------------------- IndexError Traceback (most recent call last) in 152 continue 153 speaker_list.append(f"{num} - {i.name}") --> 154 if lines[0].startswith("\ufeff"): # lamp oil, rope, boms? 155 lines[0] = lines[0][1:] 156 if lines[0].startswith("wavs/") and not os.path.exists(os.path.join(target_path, i.path, "wavs")): # not the most thorough check but IndexError: list index out of range

👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 👆 I started to activate the second stage in your Colab - hecko ✗ [epic hax'd by ubershit], but I don't understand why it's not working for me. Perhaps I should not put my audio files with the list.txt file into a folder and then that folder into an archive, but instead just put the files into an archive? Or does it not matter in terms of errors?

in a list file, I rewrote what I heard in each of my audio files. But when I watched the video tutorial , and saw in the subtitles how it was said that I should "answer what I hear" in each of my audio files, then this got me into a stupor. Do I really have to "answer" as if to a question? But if in 1 of my files I hear "I am the destroyer of worlds!" then what answer should I rewrite? It's strange. Therefore, I did not answer as if to a question, but simply listened to the words of each of my files, and simply rewrote them, thinking where to put a comma, and where an exclamation point, etc. Did I do it right?

@hecko hi pewk someone wants u

uhhhhhhhhhhhhhhh that is

extremely weird

😰

what type of archive is it

and what did you put in the

datasets

field

https://drive.google.com/file/d/1ZZkzKoNYqTsNnTU-m_wo92Z5IQUGveS0/view?usp=sharing i can give it to you to see

i put "/content/drive/MyDrive/Deathwing.RAR"

ohhhhhhhhhh i see the problem

because rar support needs to be installed separately

and i have a thing for that but it only does it if it detects a

.rar