(SCAM)Making a realistic female voice in real time.
Making a female voice!
The article is educational in nature, we do not call for anything and do not oblige. The information is presented for informational purposes only.
More interesting articles on our Telegram : @dcwurld
Subscribe to the channel and share the link to the article with your friends!
To begin with, we need a medium-powerful video card, on a weak video card there will be a large delay.
The program itself, download link and tutorial below:
On the site that opens, scroll down and click on Hugging face as in the screenshot:
Next, we see a bunch of files:
If you have a poppy, then download the poppy version.
If you have an AMD video card, then download the onnxDirectML version (the newest one)
If you have an Nvidia graphics card, then download the onnxgpu version (the newest one)
Unpack the archive and look for the file " start_http.bat" in the folder and run it.
You will start downloading files and eventually the panel will start:
To output sound from the program, we need a virtual audio cable (VAC) Virtual Audio Cable.
It is not necessary to unpack the archive; to install, run this file:
In the installer, we wait everywhere further and agree with everything.
IMPORTANT: when installing the cable, it will set itself as the default audio output device, do not forget to change it back to your headphones, and also do not forget to put line 1 in the input device in the discord.
I have translated the interface into Russian for you (only what we need):
In Quality we set from 192 to 512 (there is no point higher), in bitrate we set 4096.
Set the key to +12 if you're testing a female voice model and -12 if you're testing a male one.
Next, click Start and profit. The voice model works!