That's amazing - for us non-coders, if there's a way to incorporate that into an automatic install and usable via the visual interface, that would be awesome!
@@Jarods_Journey yea i saw that now, I managed to get it working with mrq-tortoise repo with deepspeed and the results are pretty good, a lot better than vits especially with rvc and fine tuned model
I have your tortoise GUI installation tutorial in a doc. but didn't you also do a separate video on how to use it? if so, please link it here or tell me the video name so I can search your channel
Hi, could you make a video about how to use tortoise tts in other languages for example polish? Because i’ve been using this tts and i must say it’s truly amazing. Keep putting in the work👍
Thanks for showing the whole process, took a while to get all this to work for me too! How would you pass Samples, Iterations and Temperature in the Fast API...or is this not possible?
Well, with hifigan, there is no samples or iterations because those are features of the diffusion model. Temperate should still be possible, though, it's buried in the code somewhere.
@@Jarods_Journey that glados-tts that I used a bit also had a hifigan vocoder, it was super fast! I am just getting started on tortoise, since the latest increase in interference time make it very interesting for real time voice generation!
I know I've got notifications on, but I keep on refreshing this channels home page, in the hope that there might be a new video.🤣 (thanks for you work)
I don't know if you have any idea of the scale of your work. I'm in Brazil and following what you've been doing. Your work is amazing. I'm going to start testing your audiobook project, where do you recommend starting? Because I saw that at one point you used the ai voice clone and at another moment you used tortoise-tts or did I miss a point?
Appreciate it, and I'm glad my work is being seen from all across the globe :)! This video is me going over tortoise-tts using the main github repo which is from neonbjb. For all of my other videos, I use the AI Voice Cloning repository from mrq. If you are going to try out the audiobook project, start with the AI Voice Cloning repository by mrq as that is what is compatible with my project. In this video, I use the main repo because hifigan has not yet been incorporated into the mrq repo.
That is very very interesting. Just wonder how good this will perform for voice cloning? Have you tried it with a fine tuned model as well? I wonder how good it will be able to clone voices if there are no reference files and it purely relies on the fine tuned weights
Thank you for your informative videos. I fallow you because you're clear and to the point. I just did another clone with your guidance. I could really use this update but I'm afraid of breaking my Tortoise. Is this as simple as doing the update or will I have all the issues you had? I don't code so it was confusing, I normally just fallow and do what a video says to do, and cut and paste code. I couldn't tell if you were having issues because you were doing the back a forth test or if the update wasn't good, or if it was just a simple update. I would try to fallow you, but I don't think this was a "how to", because you went so fast. Should I wait to update, considering my coding ignorance?
This wasn't a how to unfortunately, and did incorporate this into the authors latest repository, though, he hasn't been active. You might have to wait for a future video I do on it
Will RVC or UVR fix that distortion or breathy sounds? If not how are you resolving those? That's become my biggest issue and have had to become a quasi sound engineer to editing audio files
I don't care about voice cloning or audio quality. How do I use it just to read text? Where do I find actual docs? is that readme really all that's available? bro...
Timestamps:
What was changed? - 0:38
Debugging and getting it running on my system - 2:15
Testing Generation Speed - 9:45
The deepspeed = True, parameter doesn't work for some reason (Windows 11).
3 seconds of generations for
17 seconds audio 😮 makes it possible to have near real time text to voice conversion.
Thats huge.
That's amazing - for us non-coders, if there's a way to incorporate that into an automatic install and usable via the visual interface, that would be awesome!
Awesome work Manmay Nakhashi
I'm really looking forward to the next version of the Audiobook Maker now ;)
link?
is there going to be a cleaned-up tutorial on how to make the updates?
Great News!! Wish you could build a code which can run through RVC model superfast
Oooh, exciting. That's incredible!
I hope this gets added to the MRQ repo very soon, can you share your fork or something?
This is actually the main tortoise branch, all of the code it there
@@Jarods_Journey yea i saw that now, I managed to get it working with mrq-tortoise repo with deepspeed and the results are pretty good, a lot better than vits especially with rvc and fine tuned model
I have your tortoise GUI installation tutorial in a doc.
but didn't you also do a separate video on how to use it?
if so, please link it here or tell me the video name so I can search your channel
Hi, could you make a video about how to use tortoise tts in other languages for example polish?
Because i’ve been using this tts and i must say it’s truly amazing. Keep putting in the work👍
That's a great idea!
any news on if and when this will be added to the audio book?
Probably 3 weeks out or so, depends on somethings that I'm working on!
Thanks for showing the whole process, took a while to get all this to work for me too! How would you pass Samples, Iterations and Temperature in the Fast API...or is this not possible?
Well, with hifigan, there is no samples or iterations because those are features of the diffusion model. Temperate should still be possible, though, it's buried in the code somewhere.
@@Jarods_Journey that glados-tts that I used a bit also had a hifigan vocoder, it was super fast! I am just getting started on tortoise, since the latest increase in interference time make it very interesting for real time voice generation!
Thanks for sharing.
I know I've got notifications on, but I keep on refreshing this channels home page, in the hope that there might be a new video.🤣 (thanks for you work)
I don't know if you have any idea of the scale of your work. I'm in Brazil and following what you've been doing. Your work is amazing. I'm going to start testing your audiobook project, where do you recommend starting? Because I saw that at one point you used the ai voice clone and at another moment you used tortoise-tts or did I miss a point?
Appreciate it, and I'm glad my work is being seen from all across the globe :)! This video is me going over tortoise-tts using the main github repo which is from neonbjb. For all of my other videos, I use the AI Voice Cloning repository from mrq.
If you are going to try out the audiobook project, start with the AI Voice Cloning repository by mrq as that is what is compatible with my project.
In this video, I use the main repo because hifigan has not yet been incorporated into the mrq repo.
That is very very interesting. Just wonder how good this will perform for voice cloning? Have you tried it with a fine tuned model as well? I wonder how good it will be able to clone voices if there are no reference files and it purely relies on the fine tuned weights
It works well with custom models (I'm using one in the video). I believe you need the reference files for it to work, but haven't tried without them.
Thank you for your informative videos. I fallow you because you're clear and to the point. I just did another clone with your guidance. I could really use this update but I'm afraid of breaking my Tortoise. Is this as simple as doing the update or will I have all the issues you had? I don't code so it was confusing, I normally just fallow and do what a video says to do, and cut and paste code. I couldn't tell if you were having issues because you were doing the back a forth test or if the update wasn't good, or if it was just a simple update. I would try to fallow you, but I don't think this was a "how to", because you went so fast. Should I wait to update, considering my coding ignorance?
This wasn't a how to unfortunately, and did incorporate this into the authors latest repository, though, he hasn't been active. You might have to wait for a future video I do on it
can you explain how to take a trained voice from the MRQ ai-voice-cloning project and use it here?
@Jarods_Journey do you know a software to change the voice in a voice recording to another voice? From male to female?
Do you plan to add that to the Audiobook Maker?
Yup :)
@@Jarods_JourneyReal Time listening, huh? 👀 RVC is already quite fast, can't wait for it man.
awe! I love you @@Jarods_Journey
You’re the best Jarod :)
Hey, where can I refer the code for test.py to test api.py and api_fast.py.
Also, I want to fine tune the model. Which code should I use?
You can check out the original repo which is neonbjb's
Will RVC or UVR fix that distortion or breathy sounds? If not how are you resolving those? That's become my biggest issue and have had to become a quasi sound engineer to editing audio files
RVC and UVR both do an OKish job at filtering them out, but if there's too much distortion, RVC can't handle it.
@@Jarods_Journey appreciate the response, thank you
I don't care about voice cloning or audio quality. How do I use it just to read text? Where do I find actual docs? is that readme really all that's available? bro...
My guy opened Powershell instead of GIT Bash just to trigger people
The deepspeed = True, parameter doesn't work for some reason (Windows 11).
I'm more confused than watching the ending of Inception.
taking out the diffusion model speeds it up but it sounded awful
Amazing, Can this be intergraded into the Web UI?
If someone wants to integrate it :)!
TIL you can get rid of the annoying warning of changes by typing git stash. Hours of work saved. Thank you.
This is a huge speed increase! They're gonna need to come up with a new name, man. Tortoise just isn't accurate any more lol
it works but it generates british accent. is there a way to not generate british accent? my sample voice are already not british
My uncle works at Nintendo and he got me a 7090 ti ultimate edition
Hi I left a comment on another video of yours tryna install Tensorboard please help
You may have to check Google on what your issue is to see if anyone has run into this before
This is wild wtf...
Cool.
Wtf!!!