Uberduck

oh it doesnt need the text transcription?

oh no nevermind you do need the text transcription

i have some ~~totally legally~~ acquired mkvs of cartoons with subtitles, i should write a script to just play the video with subtitles and have the user manually sort the subtitles into which character is saying them

that could allow super quick creation of datasets

well if the cut the files not longer than 12 seconds and later convert the files into the right settings i see no problem

some people here have automated stuff too so it can work
bruh even i use tacotron2autotrim to do 50% of the job with transcribing

the whole point of using a cartoon mkv is all of the subtitles are already done and the voice clips are already trimmed

all that manually needs to be done is assign voice lines to characters

so right now im doing the voice labeling, what happens if the sentences are questions/exclamations?

actually im questioning about the questioning sentences

some of the lines i have contain 1 to 2 groans

Ignore the groans but keep the question marks and exclamation points. Punctuation marks influence intonations and pauses.

https://pieksend.github.io/posts/2021-05-09-tacotron-faq/

https://pieksend.github.io/posts/2021-06-04-how-to-make-a-good-dataset/

I am currently working on a Bill Nye model.

I found out he had like a 6 hour audiobook, but I don't have the bobux so I am using the free sample.

If they're part of an acronym and they're pronounced separately, separate them. If they're pronounced as one or more words, write the words that are closest to the intended pronunciation.