It is incredible. I saw some comments where they say that the accent is totally lost, but I tried it and the accent is the same!!!! I thank you very much for this, it is what I was looking for a long time. Its just perfect ❤
I have seen this problem with many UA-camrs who say AI clones of their voice is not matching with their original voice. The output created above is "exactly" sounding like you. But you can't realize that because we hear our voice in a different way than how others hear it!
The timbre is ok, the inflections and the accents got totally lost. Last year, this technique would have been incredibly good. Today there are much better options. AI is evolving at the speed of light.
@@engineerprompt what if we want to render long paragraph? should we cut it into sentences and render it one by one, or is there any way to render all of it at once?
Could you please make a video with fully detailed instructions on how to install Tortoise and get it working? Like, instructions for total beginners? 😅You skipped the part of the process between finding it on Github and adding a new voice, and that's the part that's the biggest mystery for me. I'd love to use voice cloning, but I've never used Python and basically all I know about it is the fact that it exists. So detailed step-by-step instructions for those of us who know nothing about coding would be very appreciated! 😅
muchas gracias estaaba buscando algo asi... excelente material y contenido del canal.. apesar que no hablo ingles muy bien y el traductor es malo sus videos se entienden y son muy claros
Awesome tutorial thanks. I was looking for a good Google Colab of Tortoise-TTS. By the way, I found your generated samples sounded just like you. It was hard to pick when you were playing them.
here's a few issues that i have: it doesn't have the same exact voice, it's lower pitched and sounds more older? it also has british accent for some odd reason. and when i try to form long sentences it starts babbling and making weird robot and inhuman noises lol
I'm asking before cloning my voice, is it possible to upload a large file and then the software will record it for me (of course with my cloning voice ) ? Thanks
Is it possible to add emotions to the generated audio? I mean after successfully cloning some voice I would like to fill up it with some emotions (exclamation, fear, sad, fading etc... like a voice actor).
The documentation says ---> you can evoke emotion by including things like "I am really sad," before your text. I've built an automated redaction system that you can use to take advantage of this. It works by attempting to redact any text in the prompt surrounded by brackets. For example, the prompt "[I am really sad,] Please feed me." will only speak the words "Please feed me" (with a sad tonality).
I don't have GPU option and have pretty powerful nVidia card installed. I only get GPU T4 and for some reason it won't accept that as an option. Has anything changed in how this works now? Does it need paid subscription?
I cannot actually hear the difference between the you and the first sample. Remember that you probably sound slightly different to yourself because you're hearing it through your body.
I haven't yet experimented but this, but great video!! Do you think it's possible to run locally and use it for a personal/local assistant on a PC? Also is there a way to get it running and working even without internet (so completely local)?
@@ecstasycheese7390 Spot on thanks a lot! I tried to install Bark on my PC last weekend but I got stuck in some dependency that just won't work. Gotta stick to Collab for the time being!
Hi, when I run the second code block I don't have the option to upload my files, and it shows "fileexistserror", I believe is something basic but I don't know what to do. Thank you
This model can only generate English but you can retrain the whole model for any other language. Here is the link with steps: github.com/neonbjb/tortoise-tts/issues/5#issuecomment-1112705908
Thanks for the video I found it really helpful. I played with it and created a synthetic voice using a Spanish speaking person as the voice model but my results were not as good as yours... is it me or the model works best with english language mainly?
Is it still necessary to chop audio up into 10 second segments at 22khz sample rate? I have many audio samples from 10 minutes to 45 minutes all at 48khz. Would these not work at all or would they cause some problem?
Hey guys. I’m just wondering - if you were able to download your voice over there and sign any note like Whitney Houston - would you use it? I mean it’s it cool to record a love song to somebody but with your own song?
Oi. É possível fazer a clonagem em português nessa área que você mostrou no tutorial? Ou teria que pegar a área de um brasileiro para eu conseguir fazer isso?
@@engineerprompt That would be fantastic. I just wonder, is there any voice model generator that works with tones and regardless to the language that's spoken??
@@engineerprompt How do I install that version if there's no installer for it???? I can't install the latest Python because Tortoise doesn't work with it, right?
Hi guys , do someone know if an AI like 11labs or anything else exist but in French ? Im french and im really looking for that but it seems impossible to find
@@desstuctorr5263 11Labs do have a model now that supports a few other languages than English. French is one of them, so it should be pretty straight forward :) Just switch the model from "Eleven Monolingual v1" to "Eleven Multilingual v1" when generating your voice lines.
google colab now sucks because you have to pay it will only let you generate 1-4 voices then ur done this is so frustating i will just do story reading without it
Make sure you run the block containing this code: # This is the text that will be spoken. text = "Thanks for reading this article. I hope you learned something." Seems like it didn't run that part.
This model can only generate English but you can retrain the whole model for any other language. Here is the link with steps: github.com/neonbjb/tortoise-tts/issues/5#issuecomment-1112705908
@@ValicsLehel That's true. Check out this repo, seems to have multilingual support, I haven't really looked into closely but probably something worth checking out: github.com/coqui-ai/TTS
@@engineerprompt I will take a look. I want to find a solution to generate my voice (or actor voices) but not the default one. Elevenlabs is ok, but cannot learn Romaniann for example. Just EN.
If you liked the video, you should check out the video on how to create your own AI Avatars here: ua-cam.com/video/V2efVSXSlqc/v-deo.html
where can i find help to install it on my anaconda pormpt window, its displaying errors
It is incredible. I saw some comments where they say that the accent is totally lost, but I tried it and the accent is the same!!!! I thank you very much for this, it is what I was looking for a long time. Its just perfect ❤
Thank you and glad you found it useful.
I have seen this problem with many UA-camrs who say AI clones of their voice is not matching with their original voice. The output created above is "exactly" sounding like you. But you can't realize that because we hear our voice in a different way than how others hear it!
The timbre is ok, the inflections and the accents got totally lost.
Last year, this technique would have been incredibly good.
Today there are much better options.
AI is evolving at the speed of light.
@@ChristianIce Can you link some of them?
Interesting point, that could actually be the case.
@Mutual induction Its absolutely free. Watch the video for number of audios :)
@@engineerprompt what if we want to render long paragraph? should we cut it into sentences and render it one by one, or is there any way to render all of it at once?
Thank youuu~~ i'm gonna use this for my favorite video game characters 😎
Always love your videos...no nonsense ... straight to the topic
Thank you!
It does sound EXACTLY like you, it's crazy. I guess it's the "hearing our own voice is weird" phenomenon that is at play here
that seems to be the case.
Not exactly, the replicated voice still has some missing accent.
@@engineerprompt I disagree it sounds as if it gave you an american accent
Could you please make a video with fully detailed instructions on how to install Tortoise and get it working? Like, instructions for total beginners? 😅You skipped the part of the process between finding it on Github and adding a new voice, and that's the part that's the biggest mystery for me. I'd love to use voice cloning, but I've never used Python and basically all I know about it is the fact that it exists. So detailed step-by-step instructions for those of us who know nothing about coding would be very appreciated! 😅
the voice sound just like your voice, minus the accent, sounds very good
Good job! It sounds like you, I know we hear own voices differently from audio recordings We hear ourselves thru our own bodies/bones
LOL. The generated voice really sounds like you. I thought it was yourself talking then I noticed you already clicked the button.
Great walk-through, thanks! BTW, the HQ sample did sound like you. It lost a bit of your accent, but overall it did sound like you.
Some others have pointed out the same. I guess, I am not used to hearing myself like that :)
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
"It lost a bit of your accent, but overall it did sound like you."
It lost all of the accent, not just 'a bit'.
muchas gracias estaaba buscando algo asi... excelente material y contenido del canal.. apesar que no hablo ingles muy bien y el traductor es malo sus videos se entienden y son muy claros
Thank you, I am glad you found it helpful. Consider subscribing to the channel, have something big planned for Spanish audience in near future 😀
"ModuleNotFoundError: No module named 'einops'" any idea how to solve this?
Mine made a crazy demon noise in the middle of a sentence.... SWEET. XD
haha, it can be unpredictable some time. Hope you had fun with it.
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
I love it bro, Thank's for sharing 👍
This tutorial is awesome! Thank you SO much!
Can you download that model to run it locally on your machine? Is the resulting file a .pth?
Awesome tutorial thanks. I was looking for a good Google Colab of Tortoise-TTS. By the way, I found your generated samples sounded just like you. It was hard to pick when you were playing them.
Great video, thanks.
Is there a limit in how much it can generate at once?
If you use it locally, then I think it will be limited by your RAM.
no ...its limited by the price of GPU spexs..hheheh
great video!
Can we use this for other languages too or is it limited to English Language, thanks in advance.
Same question I was about to ask
Thanks for sharing this. Is there anyway to make it read 1,000 words at a time?
Hey why do I always get a message "maximum stack size exceeded?"
There was a Brazilian who translated this video with the tool, LOL
😀
The clone is spot on, but a bit cleaner english accent.
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
thank you very much .. dose it work with language other than english ???
It seems the AI struggles with non-American accents
So do Americans.
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
@@fkxfkxas an Australian who went to the US for 5 weeks I can confirm this is very true🤣
It's a racist AI
It’s mostly American companies and American voices being used as examples mostly cuz it’s US English 90% of the time
here's a few issues that i have:
it doesn't have the same exact voice, it's lower pitched and sounds more older? it also has british accent for some odd reason. and when i try to form long sentences it starts babbling and making weird robot and inhuman noises lol
your last codes not working, everyhting is " not recognized as an internal or external command,
operable program or batch file."
You sir have my respect!
Thank you!
is it possible to make this text to speech work in another language?
Not with Tortoise, I think coqui supports that.
I'm asking before cloning my voice, is it possible to upload a large file and then the software will record it for me (of course with my cloning voice ) ? Thanks
Def gonna try this iut
Is it possible to add emotions to the generated audio? I mean after successfully cloning some voice I would like to fill up it with some emotions (exclamation, fear, sad, fading etc... like a voice actor).
I don't know if that's possible with any software yet. Maybe bark?
The documentation says ---> you can evoke emotion by including things like "I am really sad," before your text. I've built an automated redaction system that you can use to take advantage of this. It works by attempting to redact any text in the prompt surrounded by brackets. For example, the prompt "[I am really sad,] Please feed me." will only speak the words "Please feed me" (with a sad tonality).
niceeeeeee now do not have to use effort on reading using my woice but just using woice clone
can the segments be less or more than 10 seconds? or does it have to be exactly 10?
No, its can be much longer, I have tested it on upto 30seconds. Its based on the hardware you are using.
I don't have GPU option and have pretty powerful nVidia card installed. I only get GPU T4 and for some reason it won't accept that as an option. Has anything changed in how this works now? Does it need paid subscription?
Can it run in a low end pc?
I cannot actually hear the difference between the you and the first sample. Remember that you probably sound slightly different to yourself because you're hearing it through your body.
Thanks, this is extremely useful; I suppose that one can feed this algorithm a file so that it can read it and output a recording?
Yes, there is a text variable in the notebook. Assign the text to it and it will do the rest.
@@engineerprompt Can you elaborate on this? I am trying to have may sentences read, but it says its too long
Can I use this for other languages except english?
I haven't yet experimented but this, but great video!!
Do you think it's possible to run locally and use it for a personal/local assistant on a PC?
Also is there a way to get it running and working even without internet (so completely local)?
thanks for the video. Have you tried bark? Looking for voice cloning model that I can train longer locally for better results. Thanks again
I get this error after running the "generate speech" cell: "NameError: name 'text' is not defined". Anyone???
Play the "# This is the text that will be spoken." box first before going down to the "# Generate speech with the custotm voice." box
@@ecstasycheese7390 Spot on thanks a lot! I tried to install Bark on my PC last weekend but I got stuck in some dependency that just won't work. Gotta stick to Collab for the time being!
Thank you i will try then tell you
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
@@saifshaikh8828 I didn't test it
Great found feature ! But is it possible to use this clone voice to speak in another language with the same accent ? 😮
Check out the latest video on thr topic
hello. i tried mine but it doesn’t play my text but the default text that comes in the colab. please any fix?
Hi, when I run the second code block I don't have the option to upload my files, and it shows "fileexistserror", I believe is something basic but I don't know what to do. Thank you
I wanted to try this. Allas, it is taking months to upload the audio files. Then I upgraded my colab to use a pro version. No success.
how do i fix this unterminated string literal (detected at line 4) when i write my text
thank you! Do you think this could work if i use other language? or you can upload other language models or somethig?
This specific one only supports english but check out github.com/coqui-ai/TTS for multilanguage support. Hope this helps
@@engineerprompt Thanks!!!
i keep getting load_voice' is not defined
im having trouble installing the tortoise, can anyone here point me to a place where i can find help
I’m getting “Maximum call stack size exceeded.” On the second last cell
Worked great until I used up the days GPU allocation!. Nice work, thanks for the effort and video.
Glad it was useful :)
Great product showcase, can I use this to generate a 1000 word text to speech in 1 go?
With local installation, probably yes.
@Prompt Engineering Do you have a video on installing this locally? I'm struggling to get it setup.
bro it doesn work cus my ram is too high and it doesnt allow me to do it again after my first try on all my google accounts :(
NameError: name 'load_voice' is not defined
How can I fix it?
great tutorial thank you
Thank you.
Does this model work well with languages other than English such as: Japanese , Chinese , Vienamese
This model can only generate English but you can retrain the whole model for any other language. Here is the link with steps: github.com/neonbjb/tortoise-tts/issues/5#issuecomment-1112705908
Does it work only in english ?
If I want to make it better in a certain language, do i need to only add more and more samples ?
This one is limited to English
does this work with other languages ?
no
Thanks for the video I found it really helpful. I played with it and created a synthetic voice using a Spanish speaking person as the voice model but my results were not as good as yours... is it me or the model works best with english language mainly?
This specific model is tailored towards English language.
@@engineerprompt small question. Can you tune fine the model with the files you give it in different runs?
Can you do this on phone
7:33 - Best bit
Is there a way to use it on a iPad Pro?
Do i have to reupload audio everyime i use it?
Is it still necessary to chop audio up into 10 second segments at 22khz sample rate? I have many audio samples from 10 minutes to 45 minutes all at 48khz. Would these not work at all or would they cause some problem?
They will work, but you will need good hardware to run it though!
Hey guys. I’m just wondering - if you were able to download your voice over there and sign any note like Whitney Houston - would you use it?
I mean it’s it cool to record a love song to somebody but with your own song?
easy
Oi. É possível fazer a clonagem em português nessa área que você mostrou no tutorial? Ou teria que pegar a área de um brasileiro para eu conseguir fazer isso?
This specific model works only with English. There are some other models that I can explore.
@@engineerprompt That would be fantastic.
I just wonder, is there any voice model generator that works with tones and regardless to the language that's spoken??
How make translate from portuguese in this?
i keep getting an error every time I try run it
Is there a limit on how many words can be spoken?
Make a video for specific person voice cloning for hindi language using google colab..
Are other languages supported?
can i clone another language and make someone speak it
How long passage it will take to cloned the voice? I wanna create the big passage in my voice, is it possible with it? Passage of At least 5000 words
You will probably have to divide the passage into different parts and then feed that into the model. That's the best way to do it.
Can we use different languages?
Is it work with other languages such as Arabic
I get this when i try to upload my audio files "MessageError: RangeError: Maximum call stack size exceeded."
Any ideas how to solve this?
I haven't faced but make sure you have enough space on your google drive and have stable internet connection.
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
Can it clone other than english voice?
Not with this version. There are other packages that can do it.
It's support any language like kannada, or only English?
This specific one, only supports English.
How to train on other language ?
The only problem is that the software doesn't pick up your accent.
Tortoise-tts also has a fine-tuning via a fork
Would love to have a look at it, any resources you recommend?
@@engineerprompt I have put a link in the description of my lastest vids to MRQs repo
@@SyntheticVoices thanks, I will check it out!
bro i am getting error
How can improve spanish?
Bro I find a problem to # Imports used through the rest of the notebook.
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
Sounds exactly like you.
Thanks, others have pointed out the same. Seems like I don't recognize my own voice 😉
It Support Arabic Voices?
Please what is the python version, I keep scipy installation error
Python 3.9.16 (in google colab)
@@engineerprompt thank you very much
@@engineerprompt How do I install that version if there's no installer for it???? I can't install the latest Python because Tortoise doesn't work with it, right?
Great Great Ai thank you very much, do you know if it works for other languages ? french ?
I think currently it works for English only but if you can collect data, you can retrain the model on other languages.
Why does it have to be 10 second segments?
You can provide it longer segments as well but its just about the compute resources
this is so COOOOL
Thanks :)
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
cool!
Hey brother, is it safe to install in my pc ? As my windows showing harmful and dangerous risk?
Hi guys , do someone know if an AI like 11labs or anything else exist but in French ? Im french and im really looking for that but it seems impossible to find
check the next video :)
@@desstuctorr5263 11Labs do have a model now that supports a few other languages than English. French is one of them, so it should be pretty straight forward :) Just switch the model from "Eleven Monolingual v1" to "Eleven Multilingual v1" when generating your voice lines.
@@Aaliyashi Yeah I already tried it.
The voice cloning is ok but it make a canadian accent that is pretty annoying
@@desstuctorr5263 Oh I see, that's a shame. At least it seems like it's something they're working on.
@@Aaliyashi Yep .
And its only an experimental version after all !
google colab now sucks because you have to pay it will only let you generate 1-4 voices then ur done this is so frustating i will just do story reading without it
Use text to speech Can videos be monetized?
why not? watch this video to learn about Google's policy: ua-cam.com/video/VjphDyQhlW8/v-deo.html
NameError: name 'text' is not defined. I followed you in every step. :(
Make sure you run the block containing this code:
# This is the text that will be spoken.
text = "Thanks for reading this article. I hope you learned something."
Seems like it didn't run that part.
@@engineerprompt was having same problem and this is the solution
Can be trained other then EN language?
This model can only generate English but you can retrain the whole model for any other language. Here is the link with steps: github.com/neonbjb/tortoise-tts/issues/5#issuecomment-1112705908
@@engineerprompt That is not easy at all :-)
@@ValicsLehel That's true. Check out this repo, seems to have multilingual support, I haven't really looked into closely but probably something worth checking out: github.com/coqui-ai/TTS
@@engineerprompt I will take a look. I want to find a solution to generate my voice (or actor voices) but not the default one. Elevenlabs is ok, but cannot learn Romaniann for example. Just EN.