How to setup the BEST Free Text-to-Speech Locally 💥 Kokoro TTS Local Installation 💥

Kokura TTS a new 82M Parameter model is the best open source TTS at the time of the recording and here we will learn how to run this!

🔗 Links 🔗

Code used in the video

❤️ If you want to support the channel ❤️
Support here:

🧭 Follow me on 🧭

[Haidnt-ch]

Thank you for the very detailed tutorial video! I honestly didn’t expect there to be so many steps. I’ll need to rewatch the video a few more times before I can follow along @@

[AIFuzz]

I made a custom GUI where you can use Orpheus locally. It makes things so much easier!

[feritkardal]

Hello, firstly thanks for the effort! I want to ask you, can we do very long texts in just one run? Is that possible? I really appreciate if you can answer back quickly, i have a hurry 🙏

[ReallySquishyHead]

Quick note: you have to install lfs twice. The first time through homebrew. The second time in git itself. I overlooked this nuance, which lead to having to clone the repo twice.

Overall it's a good tutorial.

[drgutman]

I managed to install it (for the windows users or if you get an error that espeak-ng not installed there's a fix on the huggingface page under "Usage" code snippet). It's fast and it has some pretty good voices, but I'm not sure that it has sentiment tokens.
Also, on the huggingface space it has a tab for longform but it's for v0.19 with speed selection. Do you know where I can see how that it is done? Like, are there some parameters that can be added on runtime?

[NotRayan]

Can I use these voices in my yt videos Without any reuse and copyright issue ?

[KarimBenabd]

Before even watching the video, I thank you for this gem. You got a new subscriber.
Any plans a video about How to setup the BEST Free Transcriber locally (Audio to text)?

[existentialcurrent]

Can we do 30 minutes of audio with this setup? If yes, awesome! If not, how much audio can we generate at once?

[xtaltheo]

Thx for the tutorial, brother! It was hard to follow along as a complete beginner, but looks like I was able to get it running. Appreciate it! I have a question, though. Can this be updated to v0.23?

[alx]

They're plenty of various tts models. I rememver was playing with cocui-tts - there were several dozens of compatible models. But this one looks good as well. Just in case, the very first (longer) generation example didn't appear in video - you played the second one twice

[muhammadhilal]

Great video. But the HuggingFace Code is updated, and the source-code provided by you is not working.

[Rajesh]

If large files are mandatory, what the use of skipping at first and then downloading again

[shriniketacharya]

Thanks for the video. was able to make it work locally on my M1 Mac but the total time it took was ~1 minute for my text. Is there a way to reduce that time?

[puneet]

Great 👍
How do i switch my default TTS on my Mac to this?
For example, I wanna make it read a PDF/books/browser pages.

[Auroria]

Best kokoro installing video. thank you, I saw that Kokoro released new version. can you do another video about that? thanks

[nkalra]

is there any difference in output of onx version and pth version?

[jeevankumarreddyravuru]

This method is not working, there is no kokoro.py file after cloning the repo, can anyone help me with zip of the old repo

[Pregidth]

Can we discuss how ethical it is to take code from github, change it slightly and then sell as an app to those who don't know? Then giving back a part of the benefit unasked to the original coder? Is there a concept for this?

[mrrubel]

One question, please answer.
From hugging face page, we can generate text 2 audio, why should we install it local machine?

[Versedian]

Seems this is already outdated, thanks your your effort. Great guide and was easy to follow. Seems from commit history, models, plbert, kokoro, and others were all deleted 12 days ago. Seems there is a "pip install kokoro" now.