ultra singer program that add lycrics and pictch

[d4ftchunk]

Any one treid using this program for Python or know of a program that works?


https://github.com/rakuri255/UltraSinge ... e#-use-gpu

[german01]

yes it works,

its not perfect, but it helps save a lot work and generates a big proportion of the song, which then must be corrected.

i use it on windows with gitbash like this:
Install gitbash https://gitforwindows.org/
Install Python 3.10 https://www.python.org/downloads/release/python-3100/
press windows button and type "gitbash", then enter

Código: Seleccionar todo

# installation
mdkir c:/projects
cd c:/projects
git clone https://github.com/rakuri255/UltraSinger.git
cd UltraSinger
source .venv/Scripts/activate
pip install -r requirements.txt
pip3 install torch==2.0.1+cu117 torchvision==0.15.2+cu117 torchaudio==2.0.2+cu117 --index-url https://download.pytorch.org/whl/cu117
cd src
# actual call of the tool
py UltraSinger.py --crepe full --whisper large-v2 -i "input/song_to_parse.mp3" --language=en

For the next run, when its already installed

Código: Seleccionar todo

c:/projects/UltraSinger
source .venv/Scripts/activate
cd src
py UltraSinger.py --crepe full --whisper large-v2 -i "input/song_to_parse.mp3" --language=en

It should create input/output/song_to_parse dir with the files in it.
Then you can correct problems with Yass or smth.

When the mp3 has foreign characters in the mp3 Tags, decription or title it tends to produce problems.
so better strip the mp3 tags out completly, and give the file a very simple name.

Also the repitch function on existing files is very nice.
You can create the text and timing in yass and then just repitch it like this:

Código: Seleccionar todo

py UltraSinger.py -i "input/songtextfile.txt" --crepe full --crepe_step_size 1

its a great tool!

hope this helps someone!

[gouthamma]

me indica que py no lo reconoce como un comando, puedes hacer un video?

[german01]

i think i might had to add the python dir to the system path.
it should be something like this on the windows commandline.

Código: Seleccionar todo

setx PATH "%PATH%;C:\Python310"

I did ask chatgpt on this, so i sadly dont remember the exact step which worked, only that the path was missing.

Also you might be able to call it by just using "python" or "python3" instead of "py" at the beginning of the commands.

And i just saw that i had to install ffmpeg and put the executables it in the python310 dir for the media parsing to work
https://ffmpeg.org/download.html#build-windows

If you put it there, and if its found by the "which ffmpeg" command it should work.

Código: Seleccionar todo

$ which ffmpeg
#expected output
/c/Python310/ffmpeg

[Flipsyde24]

When i run this code :
"# installation
mdkir c:/projects
cd c:/projects
git clone https://github.com/rakuri255/UltraSinger.git
cd UltraSinger
source .venv/Scripts/activate
pip install -r requirements.txt
pip3 install torch==2.0.1+cu117 torchvision==0.15.2+cu117 torchaudio==2.0.2+cu117 --index-url https://download.pytorch.org/whl/cu117
cd src
# actual call of the tool
py UltraSinger.py --crepe full --whisper large-v2 -i "input/song_to_parse.mp3" --language=en"

I have the following message :
"bash: .venv/Scripts/activate: No such file or directory
bash: pip: command not found
bash: pip3: command not found
bash: py: command not found"

Can anyone help?