# machine-learning
- u
user
09/13/2021, 12:42 AMThis all seems incredibly overbearing- - z
Zephyr
09/13/2021, 12:44 AMYou'll get the hang of it. - a
Alexius08
09/13/2021, 10:02 AMOnly 80 voice clips at a total of almost 11 minutes, and it went okay. - c
cengi
09/13/2021, 11:52 AMI reached my limit on colab. Can I continue the training on another google account ? - h
hecko
09/13/2021, 1:08 PMyeah absolutely just make sure to pick your main account when logging into google drive, that way it'll access the model file in there - c
cengi
09/13/2021, 1:12 PMOh i thought i needed to transfer. Thanks - a
Alexius08
09/13/2021, 2:48 PMRetraining my Carl Sagan TTS model since it can't even handle the Navy Seal copypasta. Increased the dataset to almost 17 minutes. - a
Alexius08
09/13/2021, 4:20 PMThis sounds much better now. If this passes more checks, I'll upload the model as a replacement to the one I submitted earlier. - e
Echoes
09/13/2021, 5:53 PMStonks on the first Epoch? Satisfactory. - t
The Wiggles and High School D×D
09/13/2021, 10:56 PMSo I'm trying the synthesis thing again, reading the instructions careully - t
The Wiggles and High School D×D
09/13/2021, 10:57 PMI guess what I need to do is properly have a LibriSpeech/TTS/ directory but I have no idea what to do for that. I remember last time I ran it that it was supposed to be metadata.xml but I could be wrong. - t
The Wiggles and High School D×D
09/13/2021, 10:57 PMI keep being told that I'm making an assertion error - t
The Wiggles and High School D×D
09/13/2021, 10:57 PMAnd I assume I'm supposed to be linking my audio clips, but there are so many that I can't imagine how I can move every single one into a TTS folder - t
The Wiggles and High School D×D
09/13/2021, 10:58 PMUnless that's not what I'm meant to do and it's something else - t
The Wiggles and High School D×D
09/13/2021, 10:58 PMGah, it's all so frustrating because I can't figure out what I need to do - c
ColdFir3
09/13/2021, 11:13 PMHow many wav files are you using 💀 - e
Echoes
09/13/2021, 11:14 PM114 i think - c
ColdFir3
09/13/2021, 11:15 PM114 wav files with stonks like that first run is legendary, did you change any of the settings away from default? - e
Echoes
09/13/2021, 11:16 PMI changed my batch size to like 1 i think - c
ColdFir3
09/13/2021, 11:16 PMOh damn, what does the batch size do/mean again - e
Echoes
09/13/2021, 11:17 PMidk, afaik it can reduce ram usage and prevent CUDA from running out of ram - e
Echoes
09/13/2021, 11:17 PMalso you know you can turn off pings when you reply right? - c
ColdFir3
09/13/2021, 11:19 PMOhh yeahh that makes sense and nahh I didn't know I dont really use discord like that to be messing with settings etc - c
cengi
09/14/2021, 12:34 AMGuys how long should the duration be of each wav? tacotron - z
Zephyr
09/14/2021, 12:35 AMYou only have 1 wav file? 🤨 - c
cengi
09/14/2021, 12:50 AMno for each wav. - c
cengi
09/14/2021, 12:51 AMI have about 100 wavs but only 8:11 minutes combined - c
cengi
09/14/2021, 1:00 PMbump - h
hecko
09/14/2021, 1:31 PM7 seconds is a good average, it might be good to vary it so the ai doesn't forget how to do short sentences - c
cengi
09/14/2021, 1:31 PMwhats the maximum?