I tried that once

Not a good idea

what happened

🥲

ptsd

There are 250 lines of voice that louis from L4D has, I don't know if it will work

dataset made by @User

Are you talking about my voice model?

I've fixed some things for you, like 16 bit PCM, Mono and 22050hz audio

with switch plus:audio converter by NCH Software

why should that work?

idk

that's what I wanted to find out

apparently justinjohn tried it and is traumatised

you know what a transcript is right?

yes?

if the text isn't what the audio says it isn't a transcript 🤦

ok call it a filelist then, doesn't matter

you need a transcript

thats the whole point

not just some random audio with some random text

I don't think you got my point

then explain your point

I can't find another way to interpret this

Machine learns that each sound corresponds to each text as a joke/experiment, create a dataset that: - has a variety of sentences in the filelist - has the same sentence in every audio file to see if the ai will produce the same sound regardless of what text you put in for synthesis

did you read the message under it

I do understand how this is meant to work

I already made my own model

yes

and?