I watched many videos about F5-TTS on youtube. You are the only one who can clearly compare the original sound and clone sound in a clear manner to the watcher. Keep up the good work!
*Great Video , thanks, Try dubbing a short documentary and import a deep voice, let's see what we can do with all available AI tools & colabs at this moment*
Your videos bring a lot of value to the community and are worthy of 1 subscription, 1 like, and 1 comment. I wish you good health and make many valuable videos for everyone to learn and follow.
I have already tried everything in the video and it is indeed very easy to use, the AI has its problems but I guess it will improve over time, the part of cloning the voices works 100 out of 10, I managed to confuse a friend with his own voice speaking in another language which was very funny. Thank you very much for the video and for taking the time to respond to my first comment
Use the format {emotion_name} your_text. For example: If the emotion is "happy": {happy} I won a prize. For multiple emotions: {happy} I'm happy. {angry} I'm angry. {sad} I'm sad. There’s no set order. Just indicate the needed emotion in curly braces before each sentence, like {emotion} your_text. Make sure you label those reference audio files the same as your emotion_name.
Thank you for the video! It's really helpful! 🙌How can i use it with another model? like, i want to try with "F5-TTS-Turkish". how can i add it properly
Search on Google to find out if someone has trained an F5TTS model for the Turkish language or train your own model. To learn how to train in different languages watch this video: th-cam.com/video/UO4usaOojys/w-d-xo.htmlsi=uzMKfs6sdDloKU9a
@@neuralfalcon There actually is a Turkish language model. I meant to ask how can I use that with your code, since it doesn't have custom model option in ui
@@QHawk7 github.com/NeuralFalconYT/F5-TTS-Demo/blob/main/F5_TTS_Latest.ipynb You may need to run: !sudo apt install ffmpeg Ensure you are connected to a GPU runtime. You may also need to install torch if PyTorch is not pre-installed on Kaggle by default. github.com/SWivid/F5-TTS
is there a way to speak with our voice and make a transfer to this voice to reproduce the emotions of tones you know let's imagine that I play the game of an angry person can the cloned voice reproduce this angry voice ?
Easy, Record a short, 15-second audio clip where you speak in a specific tone, like angry, sad, or happy. Use this audio as a reference in F5 TTS, and the output voice will match your chosen emotion, such as anger.
@@neuralfalcon thank you, it worked. Now I have a checkpoint file trained through TTS-F5 but I don't know where to inference through, can you help me, I need python script
Yes, but you either need to train the model in French yourself or wait for someone else to do it. The best option right now is to pay for a service like ElevenLabs.io to clone your voice.
It only supports English and Chinese , but you can train it in other languages. Watch this video to learn how: th-cam.com/video/GmketyZW2c4/w-d-xo.html
github.com/NeuralFalconYT/F5-TTS-Demo/blob/main/F5-TTS-lightning-ai.ipynb Download this notebook and upload it to lightning.ai/. Make sure to switch to GPU.
It only supports English and Chinese , but you can train it in other languages. Watch this video to learn how: th-cam.com/video/GmketyZW2c4/w-d-xo.html
If you used a virtual environment delete the f5-tts folder and clean windows cache folder. or, if you used just pip to install f5 tts pip uninstall F5-TTS and clean windows cache folder.
Why did the page ask me for permission to use my microphone? Do not enter the pinned link, you will probably be hacked... The video seemed useful but better not risk it
Thank you for your comment! It sounds like you might not be familiar with how Gradio applications work. The page requests microphone permission because the app needs to record or upload audio in order to clone it. Our code prioritizes recording audio before launching the app, which is why microphone access is required. If you're interested, you can learn more about this in the Gradio documentation here: www.gradio.app/docs/gradio/audio .
@@weini-sf3pu first you need a GPU to use 'nvidia-smi' then if you are running in a jupyter notebook '!nvidia-smi' Or if you are running in terminal just 'nvidia-smi'. Else you can skip this. Use another way to find the cuda version to install the pytorch .
Happy 1k subs
Thank you bro
I watched many videos about F5-TTS on youtube. You are the only one who can clearly compare the original sound and clone sound in a clear manner to the watcher. Keep up the good work!
Glad I could help!
thank you thank you thank youuu.
Neural Falcon you saved me a lot of time THANK YOU SO MUCH
Glad it helped!
Thank you very much and especially for your patience
*Great Video , thanks, Try dubbing a short documentary and import a deep voice, let's see what we can do with all available AI tools & colabs at this moment*
Your videos bring a lot of value to the community and are worthy of 1 subscription, 1 like, and 1 comment. I wish you good health and make many valuable videos for everyone to learn and follow.
Thank you so much man! You're awesome!
I have already tried everything in the video and it is indeed very easy to use, the AI has its problems but I guess it will improve over time, the part of cloning the voices works 100 out of 10, I managed to confuse a friend with his own voice speaking in another language which was very funny.
Thank you very much for the video and for taking the time to respond to my first comment
thanks for the tutorial
Happy to help
2nd audio and first model super
Pls make a video on how to use multi-speech option of this model, I'm having troubles using it
11:18 watch this video th-cam.com/video/6i0cXSvyz98/w-d-xo.htmlsi=IZ8FKfAD7l0sqmgV
Use the format {emotion_name} your_text.
For example:
If the emotion is "happy": {happy} I won a prize.
For multiple emotions: {happy} I'm happy. {angry} I'm angry. {sad} I'm sad.
There’s no set order. Just indicate the needed emotion in curly braces before each sentence, like {emotion} your_text.
Make sure you label those reference audio files the same as your emotion_name.
@neuralfalcon thx for helping me out
Excellent!
Can you make a colab version for Fish Speech tts?
Someone already did it. Try this 😀
colab.research.google.com/drive/1trBvrdgyI-Ntd45ZnlT5lhGsI_HnKjC1?usp=sharing
@@neuralfalconI tried that one, but I can't seem to make it work. Could you...make a tutorial on how to run it? 🥺🙏
@@gg69155 i will try
@@neuralfalconthank you 😊
Do you have the same but for lipsyn? Something on google collab or similar? I want to lipsync audio and a video, not an image.
Yes, we have Wav2Lip.
github.com/Rudrabha/Wav2Lip
My google Colab link:
github.com/NeuralFalconYT/wav2lip
I've got an error in second step saying -> name 'base_path' is not defined
Because, you forgot to run the cell above, where base_path = "/content".
Run the cell above first, then run the next one afterward.
2:57 That's Dva right?
Yes
Can you make a video about how to train your model in a different language, please?
github.com/SWivid/F5-TTS/discussions/143
th-cam.com/video/RQXHKO5F9hg/w-d-xo.html
Watch this video: th-cam.com/video/GmketyZW2c4/w-d-xo.html
Watch this video : th-cam.com/video/UO4usaOojys/w-d-xo.html
Thank you for the video! It's really helpful! 🙌How can i use it with another model? like, i want to try with "F5-TTS-Turkish". how can i add it properly
Search on Google to find out if someone has trained an F5TTS model for the Turkish language or train your own model.
To learn how to train in different languages watch this video:
th-cam.com/video/UO4usaOojys/w-d-xo.htmlsi=uzMKfs6sdDloKU9a
@@neuralfalcon There actually is a Turkish language model. I meant to ask how can I use that with your code, since it doesn't have custom model option in ui
Can I get this to work on kaggle?
Yes
@@neuralfalcon
How?
@@QHawk7
github.com/NeuralFalconYT/F5-TTS-Demo/blob/main/F5_TTS_Latest.ipynb
You may need to run:
!sudo apt install ffmpeg
Ensure you are connected to a GPU runtime.
You may also need to install torch if PyTorch is not pre-installed on Kaggle by default.
github.com/SWivid/F5-TTS
is there a way to speak with our voice and make a transfer to this voice to reproduce the emotions of tones you know
let's imagine that I play the game of an angry person can the cloned voice reproduce this angry voice ?
Easy, Record a short, 15-second audio clip where you speak in a specific tone, like angry, sad, or happy. Use this audio as a reference in F5 TTS, and the output voice will match your chosen emotion, such as anger.
When I do it, it says
"No module named onx"
yeap, new bug
can i training with own voice, for example: optimus voice..
Yes 100%, copy this notebook and use F5-TTS colab.research.google.com/github/NeuralFalconYT/F5-TTS-Demo/blob/main/F5_TTS_Latest.ipynb
I checked your collab, and it doesn't work
It's working
@@neuralfalcon thank you, it worked. Now I have a checkpoint file trained through TTS-F5 but I don't know where to inference through, can you help me, I need python script
Link no longer works. Any new links?
colab.research.google.com/github/NeuralFalconYT/F5-TTS-Demo/blob/main/F5_TTS_Latest.ipynb
Or follow official instructions:
github.com/SWivid/F5-TTS
It was PERFECT for that window though and I thank you so much.
Hello, can i also clone a voice in french?😊
Yes, but you either need to train the model in French yourself or wait for someone else to do it.
The best option right now is to pay for a service like ElevenLabs.io to clone your voice.
huggingface.co/RASPIAUDIO/F5-French-MixedSpeakers-reduced
can we dubbed into other language like from english to hindi
It only supports English and Chinese , but you can train it in other languages. Watch this video to learn how:
th-cam.com/video/GmketyZW2c4/w-d-xo.html
huggingface.co/SPRINGLab/F5-Hindi-24KHz
F5-TTS Hindi: th-cam.com/video/Pb3Zx562Juw/w-d-xo.html
I tried, is it possible to make him speak with a French accent, he still has difficulty or can I speak to the creator to ask him the question?
th-cam.com/video/RQXHKO5F9hg/w-d-xo.html
Watch this video : th-cam.com/video/UO4usaOojys/w-d-xo.html
huggingface.co/RASPIAUDIO/F5-French-MixedSpeakers-reduced
Can you please make it work on LightningAI bro?
github.com/NeuralFalconYT/F5-TTS-Demo/blob/main/F5-TTS-lightning-ai.ipynb
Download this notebook and upload it to lightning.ai/. Make sure to switch to GPU.
HEY.. IS THE PROCESS FOR E2 IS SAME?
Yes same, just choose E2-TTS button
how to install F5-TTS bro
pip install git+github.com/SWivid/F5-TTS.git
f5-tts_infer-gradio
For English: colab.research.google.com/github/NeuralFalconYT/F5-TTS-Demo/blob/main/F5_TTS_Latest.ipynb#scrollTo=dB3clv5KWtQG
For Hindi:
colab.research.google.com/github/NeuralFalconYT/F5-TTS-Demo/blob/main/F5_TTS_Hindi_Small.ipynb
F5 tts Can Support Hindi voice Give Answer ?
It only supports English and Chinese , but you can train it in other languages. Watch this video to learn how:
th-cam.com/video/GmketyZW2c4/w-d-xo.html
@@neuralfalcon Please you can make video realistic hindi tts voice
Watch this video : th-cam.com/video/UO4usaOojys/w-d-xo.html
huggingface.co/SPRINGLab/F5-Hindi-24KHz
F5-TTS Hindi: th-cam.com/video/Pb3Zx562Juw/w-d-xo.html
Làm sao để xóa chúng khỏi máy tính vậy?
If you used a virtual environment delete the f5-tts folder and clean windows cache folder.
or, if you used just pip to install f5 tts
pip uninstall F5-TTS
and clean windows cache folder.
Why did the page ask me for permission to use my microphone? Do not enter the pinned link, you will probably be hacked... The video seemed useful but better not risk it
Thank you for your comment! It sounds like you might not be familiar with how Gradio applications work. The page requests microphone permission because the app needs to record or upload audio in order to clone it. Our code prioritizes recording audio before launching the app, which is why microphone access is required. If you're interested, you can learn more about this in the Gradio documentation here: www.gradio.app/docs/gradio/audio .
@@neuralfalcon Thank you very much for replying to my comment, I will read the documentation, it is true that I am not familiar with the application
does it work in indonesian language?🙏
For now it's a big 'NO' . You need to train on Indonesian language from scratch. You can use elevenlabs but it's paid.
Watch this video: th-cam.com/video/GmketyZW2c4/w-d-xo.html
Watch this video : th-cam.com/video/UO4usaOojys/w-d-xo.html
Will it work with hindi language
For now it's a big No, you need to train for other languages From scratch
@@neuralfalcon do you know how to train it? Or any reference video of it?
@@abhishekkumar-bz1ql github.com/SWivid/F5-TTS/issues/87
Watch this video : th-cam.com/video/UO4usaOojys/w-d-xo.html
F5-TTS Hindi: th-cam.com/video/Pb3Zx562Juw/w-d-xo.html
*Is it Multi-language?*
Only English and Chinese
Watch this video : th-cam.com/video/UO4usaOojys/w-d-xo.html
when use Generate TTS, get an error " FileNotFoundError: [Errno 2] No such file or directory: 'nvidia-smi' ", can you help me ?
@@weini-sf3pu yes send screenshot at NeuralFalcon@proton.me
@@weini-sf3pu first you need a GPU to use 'nvidia-smi' then if you are running in a jupyter notebook '!nvidia-smi'
Or if you are running in terminal just 'nvidia-smi'. Else you can skip this.
Use another way to find the cuda version to install the pytorch .
uncheck remove silence
You can uncheck the silence option, which may create silence in the generated audio .