00:01 Learn to change your voice in real time for Discord. 01:33 Extract audio from video and clean it. 03:30 Using the AI Live Voice Changer tool for real-time voice manipulation. 05:22 Adjust batch size if encountering memory issues 07:13 How to use Discord Voice Changer on Windows with Nvidia GPU 09:06 Adjust volume, GPU settings, and algorithm for best voice quality. 11:00 Using AI Live Voice Changer in Discord 13:03 Fixing crackling voice in Discord
Dude thank you so much. You explained everything else a lot better than the other people I’ve watched. Really appreciate this video, it saved me a lot of headaches. Can’t thank you enough.
Thanks for watching and for your feedback! Yeah I’m not happy with the way the voice sounds currently. Later I’m going to train it again with different settings.
@@AiVOICETUTOR Some voice are imposible to train even with high quality audio, why ? i mean the sound is not even close to the voice even with 300 epoch
Yeah I've noticed that as well. Some voice sound more like the original voices compared to others. Maybe someone else can explain it but I have no idea why that is. I only hope that we'll soon get an improved version of RVC (maybe v3)
Just found your channel - thanks for content and knowledge (text2voice space is quite messy and not well established - yet). I see you use different GUI now compare to previous tutorial that you made month ago + you mentioned RVC GUI - once again, single channel and already 3 options. So in terms of best possible quality we need head to RVC GUI - correct? (to create pth file), what about best possible quality for text2voice ?
Glad you found the channel and that you like it! Good question but it all depends what your goal is. The three videos have their separate purposes. One is for training (RVC Beta) and cloning your own voice in non real-time (with RVC GUI) which IMO gives the best results overall. The second video was about Text-to-speech which uses a different tool (TortoiseTTS) and is not compatible to the models trained in RVC Beta. AFAIK this the best free option with highest quality to use text to voice that is currently available.The third video is about being able to train a voice (in RVC Beta again) which can be generated in real-time (with Real-Time Voice Changer by W-Okada) so it can be used to voice chat. Hope this makes this a bit clearer.
I got this error: Traceback (most recent call last): File "D:\RVC-beta\RVC-beta0717\trainset_preprocess_pipeline_print.py", line 79, in pipeline audio = load_audio(path, self.sr) File "D:\RVC-beta\RVC-beta0717\my_utils.py", line 19, in load_audio raise RuntimeError(f"Failed to load audio: {e}") RuntimeError: Failed to load audio: module 'ffmpeg' has no attribute 'input'
From what I can tell you are using the non python version of FFmpeg. Try runnning this command in a command window: "pip install ffmpeg-python" (You might have to uninstall the other FFmpeg by running "pip uninstall ffmpeg") Hope this helps
Sorry that it didn't work. The only thing I can suggest is keeping an eye on the Huggingface site and try it with the next version that will be released there (or try it with an older version).
Hi sounds like you opened the file in windows explorer (which is showing you the contents rather than the actual 7z file). Just navigate to the downloaded 7z file and right click it to unzip (Or open 7zip and navigate to the downloaded file). Hope that helps
my voice still sounds kinda choppy.. i have a 3080 and i donwload the gpu version of MMVC but still its still kinda sounds meh for the gura voice... chunk set 512 extra 65536 F0 Harvest... trancate 300 been play around ... just doesnt sound that smooth compare to yours
Hi, sadly I don't think it will let you train a voice because you need a dedicated graphics card. Converting your voice to a pertained voice can be done with cpu only but will be slower
Hello, I've now got voice changer installed and works with the standard voices that are pre-installed. However, When i download a new voice, i dont hear anything. Its all the same settings, Do you know what the issue could be?
@@AiVOICETUTOR hey, thanks for the reply. i sorted the issue by downloaded 12 (not 12a) and converting the model to onyx and works now. apparently its due to a bug in 12b and the Quadro graphics card i have. Not sure why Quadro would affect it as it uses the same cuda library but im working 🙂
Hi when i click on Process data on Gradio in CMD.exe i have " ValueError: Invalid Literal for int() withe base 10: 'VOICE\\RVC-beta0717\\VOIX' Any tips ? Thanks
@@AiVOICETUTOR yes indeed a space in folder's name... my bad! thank you ^^ An other thing if you can help me, after clicking on train feature index i have no .pth created in " Weights " but .index is well created in the good folder... Ty sir
This seems to be a strange bug that a few people are experiencing. Does the output of the command video look the same as in the video? Like does it say "Saving model and optimizer state at epoch 300 to ..." ?
Is it possible that you are using your CPU instead of GPU? It could be a bug of the tool as you're not the only one with the issue. Maybe keep an eye on this thread: github.com/w-okada/voice-changer/issues/778
@@AiVOICETUTOR thanks for your concern dude ...just sorted all things ... Working mode on and btw you could tell people how to use that ai voice cloning (text to voice ) in google collab ...just wasted 2days running it on local maching, now i am doing that project on Google collab .. Thanks a lot dude for you content..love you from india
Thank you very much for the feedback! Glad that you sorted it out. I have not worked with Colab yet myself but will definitely look into it in the future. Greetings to India!
RVC V2 only works with NVIDIA AFAIK but for the voice changer you can download the directML version and set it to use GPU 0 as someone in the comments here suggested.
All of the voices you used just sound like you with pitch changes. Can't believe you would think they sound good at all. You could do the same thing with a lot less work.
RVC V2 only works with NVIDIA AFAIK but for the voice changer you can download the directML version and set it to use GPU 0 as someone in the comments here suggested.
00:01 Learn to change your voice in real time for Discord.
01:33 Extract audio from video and clean it.
03:30 Using the AI Live Voice Changer tool for real-time voice manipulation.
05:22 Adjust batch size if encountering memory issues
07:13 How to use Discord Voice Changer on Windows with Nvidia GPU
09:06 Adjust volume, GPU settings, and algorithm for best voice quality.
11:00 Using AI Live Voice Changer in Discord
13:03 Fixing crackling voice in Discord
Dude thank you so much. You explained everything else a lot better than the other people I’ve watched. Really appreciate this video, it saved me a lot of headaches. Can’t thank you enough.
Thanks man. That comment made my day. Really appreciate it!
Straight to the point. Instructions clear. All info in a single video. Great tutorial video, thank you.
Thank you very much! Glad you like it!
Love your videos and precisely explained. Thanks so much and looking forward to learning much more from you with your new videos.
Thank you very much for your comment! This is one of the main reasons why I love making the videos.
Thank you for this tutorial
voice sounds robotic but it's ok
Thanks for watching and for your feedback! Yeah I’m not happy with the way the voice sounds currently. Later I’m going to train it again with different settings.
@@AiVOICETUTOR Some voice are imposible to train even with high quality audio, why ? i mean the sound is not even close to the voice even with 300 epoch
i have 20 minute HQ vocal
Yeah I've noticed that as well. Some voice sound more like the original voices compared to others. Maybe someone else can explain it but I have no idea why that is. I only hope that we'll soon get an improved version of RVC (maybe v3)
Hak edenler hak ettiği yerde değiller malesef oysa ne güzel sesin var
I can’t talk through it on discord at all like it doesn’t detect the vba cable
How do you have tea time and without delays? Even with rtx 4090 I have milliseconds of delay. How to do?
Respect to you my brother
Thanks and respect to you for commenting
Just found your channel - thanks for content and knowledge (text2voice space is quite messy and not well established - yet). I see you use different GUI now compare to previous tutorial that you made month ago + you mentioned RVC GUI - once again, single channel and already 3 options.
So in terms of best possible quality we need head to RVC GUI - correct? (to create pth file), what about best possible quality for text2voice ?
Glad you found the channel and that you like it! Good question but it all depends what your goal is. The three videos have their separate purposes. One is for training (RVC Beta) and cloning your own voice in non real-time (with RVC GUI) which IMO gives the best results overall. The second video was about Text-to-speech which uses a different tool (TortoiseTTS) and is not compatible to the models trained in RVC Beta. AFAIK this the best free option with highest quality to use text to voice that is currently available.The third video is about being able to train a voice (in RVC Beta again) which can be generated in real-time (with Real-Time Voice Changer by W-Okada) so it can be used to voice chat. Hope this makes this a bit clearer.
Theres some delay when i use the live voice changer is there something im missing to make it faster?
during the final step, i get the error "the server socket has failed to listen on any local network adress" and then "runtime error" any ideas ?
I hope someone else can answer this because I can't find any info on this issue
i did everthing but W-Okada dont work for me?
Sorry it's not working for you. What exactly is the issue?
I got this error:
Traceback (most recent call last):
File "D:\RVC-beta\RVC-beta0717\trainset_preprocess_pipeline_print.py", line 79, in pipeline
audio = load_audio(path, self.sr)
File "D:\RVC-beta\RVC-beta0717\my_utils.py", line 19, in load_audio
raise RuntimeError(f"Failed to load audio: {e}")
RuntimeError: Failed to load audio: module 'ffmpeg' has no attribute 'input'
From what I can tell you are using the non python version of FFmpeg. Try runnning this command in a command window: "pip install ffmpeg-python" (You might have to uninstall the other FFmpeg by running "pip uninstall ffmpeg") Hope this helps
@@AiVOICETUTOR didn't work 😢
Sorry that it didn't work. The only thing I can suggest is keeping an eye on the Huggingface site and try it with the next version that will be released there (or try it with an older version).
Can you make a video for macbook installation? Having a lot of problems.
M1 silicon
I had some issues myself when I tried on a M1 a while ago. I hope it will get added to Pinokio (ruclips.net/video/ln1qEglnpMo/видео.html) soon.
Hey do you know why when I download the rvc-beta-7z it doesn't come out like a file but a compressed forlder which I can't extract with my 7zip?
Hi sounds like you opened the file in windows explorer (which is showing you the contents rather than the actual 7z file). Just navigate to the downloaded 7z file and right click it to unzip (Or open 7zip and navigate to the downloaded file). Hope that helps
my voice still sounds kinda choppy.. i have a 3080 and i donwload the gpu version of MMVC but still its still kinda sounds meh for the gura voice... chunk set 512 extra 65536 F0 Harvest... trancate 300 been play around ... just doesnt sound that smooth compare to yours
Someone in the comments suggested setting the tool to "Server" for better quality, Maybe you can try that?
Hello, I have a laptop of a Intel(R) Core(TM) i5-1035G7 CPU @ 1.20GHz 1.50 GHz, with Intel (R) Iris (R) Plus Graphics, does it work if I do this?
Hi, sadly I don't think it will let you train a voice because you need a dedicated graphics card. Converting your voice to a pertained voice can be done with cpu only but will be slower
Hello, I've now got voice changer installed and works with the standard voices that are pre-installed. However, When i download a new voice, i dont hear anything. Its all the same settings, Do you know what the issue could be?
Are you importing them just like in the video? And you're not getting any output at all?
@@AiVOICETUTOR hey, thanks for the reply. i sorted the issue by downloaded 12 (not 12a) and converting the model to onyx and works now. apparently its due to a bug in 12b and the Quadro graphics card i have. Not sure why Quadro would affect it as it uses the same cuda library but im working 🙂
Ahh interesting! Awesome that you figured that out and thanks for sharing the solution here!
Hi when i click on Process data on Gradio in CMD.exe i have " ValueError: Invalid Literal for int() withe base 10: 'VOICE\\RVC-beta0717\\VOIX' Any tips ? Thanks
Hi, do you have an spaces or special characters in your folder path?
@@AiVOICETUTOR yes indeed a space in folder's name... my bad! thank you ^^ An other thing if you can help me, after clicking on train feature index i have no .pth created in " Weights " but .index is well created in the good folder... Ty sir
This seems to be a strange bug that a few people are experiencing. Does the output of the command video look the same as in the video? Like does it say "Saving model and optimizer state at epoch 300 to ..." ?
@@AiVOICETUTOR Nah, it says Move model to CPU. i dont have Nvidia cards. It not building Epoch like you :/
Oh RVC Beta won't work with AMD cards properly for now.
Can this be use on vicidial or go auto dial ????
I don't know the tools you mentioned but if they freely let you select the audio sources on your PC as Mic input, then it should work.
hi sorry! i have a problem with the res, its always hovering at around 40000 ms-1. how do i fix that
Is it possible that you are using your CPU instead of GPU? It could be a bug of the tool as you're not the only one with the issue. Maybe keep an eye on this thread: github.com/w-okada/voice-changer/issues/778
am i doing something wrong? the ai voice wont work in discorrd
Maybe double check that you selected the right virtual input/output devices in RVC GUI and discord
@@AiVOICETUTOR i did
I have some doubt related to tortise video that you made to gemerate voice from text .how can i contact you
I replied to your comment over there. Hope it helps!
@@AiVOICETUTOR thanks for your concern dude ...just sorted all things ...
Working mode on and btw you could tell people how to use that ai voice cloning (text to voice ) in google collab ...just wasted 2days running it on local maching, now i am doing that project on Google collab ..
Thanks a lot dude for you content..love you from india
Thank you very much for the feedback! Glad that you sorted it out. I have not worked with Colab yet myself but will definitely look into it in the future. Greetings to India!
Does this work on AMD graphics cards? For some reason I can't see my video card.
AFAIK, both the RVC Beta and the voice changer do not work with AMD (yet?) although the voice changer can be used with CPU only
@@AiVOICETUTOR Using the direct ml version and setting it to gpu 0 makes it work on my RX 7900 XT.
Awesome! Thanks for sharing that information!
@@josh2482 How you do that vbro ?
My GPU isnt recognize, I have an amd one, what can i do?
RVC V2 only works with NVIDIA AFAIK but for the voice changer you can download the directML version and set it to use GPU 0 as someone in the comments here suggested.
Load model from pretrain/rmvpe.onnx failed:Protobuf parsing failed. error
All of the voices you used just sound like you with pitch changes. Can't believe you would think they sound good at all. You could do the same thing with a lot less work.
Anyway for amd gpu to process it?
RVC V2 only works with NVIDIA AFAIK but for the voice changer you can download the directML version and set it to use GPU 0 as someone in the comments here suggested.
Everything worked, but the cloned voice just sounded like me.
Have you tried downloading another pertained voice?
@@AiVOICETUTOR I used a file of a voice that was about 4 or 5 minutes long. Went thru all the steps, and then talked and it just sounded like me.
Does this work on MAC?
Yes the voice changer works on Mac. But you can only use the pretrained voices as RVC (used to train your own voice models) only works on windows
is this safe?
It depends on what you use it for but I doubt discord will ban you for using a voice changer
start hhtp batch file crashing
Sorry its not working for you. What's the error message?
I have a query in go web.bat it's system 32 directory and it's saying find no such file directory how to fix
Try using paths for everything which aren't too long, and don't contain any spaces or special characters.