#🔒 Local TTS via Py Advice

25 messages · Page 1 of 1 (latest)

hexed iris
#

Requirements:

  • Needs to be able to adjust with timestamps from (.str).
  • Needs to be human like

What im thinking for now is Coqui.ai but im not sure of quality. Any one that has recommendation for AI models or ideas of how to do this.

(this is for dubbing a single person for now at least)

vivid grailBOT
#

@hexed iris

Python help channel opened

Remember to:

  • Ask your Python question, not if you can ask or if there's an expert who can help.
  • Show a code sample as text (rather than a screenshot) and the error message, if you've got one.
  • Explain what you expect to happen and what actually happens.

:warning: Do not pip install anything that isn't related to your question, especially if asked to over DMs.

Closes after a period of inactivity, or when you send !close.
river finch
#

I have no idea what you mean by "adjust with timestamps" - adjust what?

hexed iris
river finch
#

eh

hexed iris
#

Like a text telling
1
00:00:00,000 --> 00:00:03,600
hi

2
00:00:04,120 --> 00:00:07,820
My name is ...

3
00:00:07,820 --> 00:00:09,900
Jhon

#

Like here it should match time so that mouth movements match

river finch
#

are you trying to build a dubbing service or something

hexed iris
#

Its something like that. I was watching anime then realized half of the episodes are not dubbed

#

So i started to do this project at my free time

#

Right now i use whisper for subtitles with timestamps also ffmpeg multiplex for dub & subtitle merging

river finch
#

I think I've seen projects that are better suited to dubbing than a simple TTS

#

but I can't be sure

#

(side note: why not just use the subtitles?)

#

you're not going to get very good dubs with AI 🙂

hexed iris
#

I thought i could use it in youtube for my friends channel by locally training their voice

#

Its just finetuning right now

river finch
#

zonos can do voice cloning too

hexed iris
#

Im gonna look in that thx very much

#

Even if time stamps dosent work i can also slow down or speed up the video so time stamps are the second priority i have at the time. So zonos looks fine.

#

for anyone looking for same thing: It looks like zonos require more work because it dosent have timestamp support or exact control over voice with timestamps. If you're willing to do some math use that because as long as how much i tested its one of best but you need to calculate the time or generate multiple versions to check (which is inefficent). Any one with other methods feel free to contribute.

river finch
#

I take it commercial tools are off the table for this?

vivid grailBOT
#
Python help channel closed for inactivity

This help channel has been closed. Feel free to create a new post in #1035199133436354600. To maximize your chances of getting a response, check out this guide on asking good questions.