# machine-learning
- c
cengi
09/14/2021, 1:31 PMmost of my stuff are below 8 seconds - h
hecko
09/14/2021, 1:58 PMthere's no hard maximum but it's best to keep it under 15 seconds, after that the ai might have trouble aligning itself to the audio and also it might not fit in the vram - c
cengi
09/14/2021, 1:58 PMThanks - t
The Wiggles and High School D×D
09/14/2021, 3:19 PMHow can I circumvent an assertion error? - h
hecko
09/14/2021, 3:20 PMthat's very generic, please send a screenshot of the entire error(s) - t
The Wiggles and High School D×D
09/14/2021, 3:46 PMPretty much no matter what directory input I give it despite it being meant to locate audio - t
The Wiggles and High School D×D
09/14/2021, 3:53 PMIt comes across as pretty vague - t
The Wiggles and High School D×D
09/14/2021, 4:05 PMI barely understand the process myself - t
The Wiggles and High School D×D
09/14/2021, 4:05 PMPutting in my dataset root tells me it will look in the folder to find LibriSpeech/Tacotron/ for the audio - t
The Wiggles and High School D×D
09/14/2021, 4:05 PMBut nothing works - t
The Wiggles and High School D×D
09/14/2021, 4:06 PMIt's an endless loop I haven't escaped for two weeks - h
hecko
09/14/2021, 4:14 PMuhhhh idk - h
hecko
09/14/2021, 4:14 PMwhat does the path look like - t
The Wiggles and High School D×D
09/14/2021, 4:24 PMDescription of the file I'm running - t
The Wiggles and High School D×D
09/14/2021, 4:24 PMAnd my path is - t
The Wiggles and High School D×D
09/14/2021, 4:24 PM!python synthesizer_preprocess_audio.py --datasets_root d/encoder/ - j
JacobLenstar
09/14/2021, 4:45 PMWendy Testaburger is not training well - h
hecko
09/14/2021, 4:46 PMhow many epochs did you do - j
JacobLenstar
09/14/2021, 4:46 PM40 - h
hecko
09/14/2021, 4:55 PMhm - h
hecko
09/14/2021, 4:55 PMidk then - h
hecko
09/14/2021, 4:55 PMmaybe make sure your transcript is lined up properly - h
hecko
09/14/2021, 4:55 PMbecause it's possible that e.g. wav 40 has the caption meant for wav 39 - t
The Wiggles and High School D×D
09/14/2021, 6:34 PMI don't know what to do The Wiki doesn't tell me anything to help me out, nor do the issues on the page - t
The Wiggles and High School D×D
09/14/2021, 6:34 PMAnd I really can't be stuck here for the rest of the month, let alone the year. I want to clone voices in two languages at once and I keep having these setbacks. - h
hecko
09/14/2021, 6:41 PM-oh i thought you were running talknet training or sth - h
hecko
09/14/2021, 6:42 PMi'mma have to look up what repo you used but in the meantime here's a thing if you wanna start over https://github.com/deterministic-algorithms-lab/Cross-Lingual-Voice-Cloning - t
The Wiggles and High School D×D
09/14/2021, 7:35 PMI used the Russian version because it has the structure done for me with set instructions for how to replace languages - t
The Wiggles and High School D×D
09/14/2021, 7:35 PMAre you sure it's worth redoing from the start? - h
hecko
09/14/2021, 7:35 PMidk