Sounds good but a bit overtrained.

yea

the dataset was a wav file 176hz

there is 0% chance yt outputting 176hz

it was 20khz

Do you mean 17khz?

Why you did even select that?

idk

Idk?

i think i was pretty sober

i dont remember anything

that i put that on 176

Can you bother to show your dataset's spectrogram?

Using Spek.

sur

um

what is spek?

is it a spectrogram viewer or smt?

spectrogram

yes

Give him a download link to it.

im installing it

u could've de noise it

so much space wasted :p

:p

train on 32k when ripping from youtube

alr

any frequency above 16khz is pretty much junk is all

o

on youtube yeah cuz they compress a lot

no wonder my other voice models sounds trash

and use uvr de noise to clean up the noise