If you want to get a better idea of what the problem could be, you could edit the start_http.bat file (it's actually just a text file) by right-click it and clicking edit (if you're on windows 11, I don't know where the edit button is but hopefully it's easy to find) and add the word pause in a new line after everything in the file, so that it looks like this:Īnd then save it and double click the start_http.bat (it still won't work, but this time it shouldn't close) and copy all the text in the terminal and reply to this comment with it and I can try to help figure out what the problem is if you want more help You need to have a dedicated gpu to use the appīut these are just some ideas I had of what could be a problem.Do you know if you have an Nvidia or an AMD gpu? If you have an AMD gpu, you have to download and use the version of the app that has the word "DirectML" in it.Two things I've thought of that could be the problem: So there could be an incompatibility with your computer that the app doesn't like Usually if it closes very quickly it means the app had a problem. Hmmm, sorry you're having problems, I tested the most recent version myself and was able to get it running. Failed to execute script 'MMVCServerSIO' due to unhandled exception: stat: path should be string, bytes, os.PathLike or integer, not NoneType TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType I can't boot it up for some reason and I've already tried using older models and they all just say this:įile "downloader/WeightDownloader.py", line 17, in downloadWeightįile "genericpath.py", line 19, in exists I changed the audio from client to server at the bottom of the app. Many buttons and settings in the app take time for it to process, so sometimes there will be a delay after changing a setting or switching a voice before it'll do it, don't worry that's normal Some of the voices that you'll find in the discord I linked above will be worse quality than others because some of them were created with lower resources, so keep in mind that some voices people share could sound much much better or much much worse than others I used a chunk of 512 and an extra of 32768 for the voice clips I recorded above Increase the "chunk" and "extra" setting for better conversion quality, but at a cost of increased latency. For Voice #1 that I recorded above I used a tune of 6, and for Voice #4 I used a tune of 12 So generally for male to female increase the pitch slider, and for female to male lower the pitch slider until you find an amount that sounds good. It's very important to adjust the tune slider in the app depending on the voice, this will change the pitch of your voice to the ai so it does a good job of matching the voice you choose. I'm not an expert but probably something like a GTX 1060 would be the minimum you'd need Things I learned that you might want to know to use the app:Īs with most other ai-powered voice changers, you'll need a computer with a decent gpu for it to run. If you see a screen like this, then select RVC If you can't even get the app to start properly, try downloading a previous version from the list of downloads (the "normal" link was renamed to "google" but it's the same thing just called differently now)Īfter you download it and unzip it, find and run the start_http.bat file (or mand if you're on mac) It will start downloading and installing everything it needs to run and then it should start Also, if it doesn't let you download because too many people are accessing it, try the "hugging_face" link directly below the "normal" link that I circled and download it from there instead There's two windows versions, if the first one doesn't work for you, try the second one. If you're having trouble finding the download, you want to find the part that looks like this, and select the one for your operating system. So use an english-speaking voice for english, for exampleīut be careful because the files people share in the ai hub discord above could contain malware when run in the voice changer There's a discord where people share voices they've made that are compatible with w-okada's voice changer (go to the voice-models channel): įinding a voice that was made in the language you plan to speak in will result in better quality and pronunciation. I found a good video of someone showcasing the app and explaining how to use it: Here are some voice clips of me using it (keep in mind the voices that come with it are not english voices and so have an accent and don't sound great in english): It's w-okada's voice changer, you can find it here: I learned about a new voice changer that might interest others that I think sounds good.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |