The vocals were removed using Ultimate Vocal Remover 5, which is an open source application.
- UVR5: github.com/Anj...
Configuration used:
- Process method: MDX-Net
- MDX-Net Model: MDX23C-InstVoc HQ
- Segment size: 512
- Overlap: 32
- GPU conversion (checked)
- Instrumental Only (checked)
This video was made with FFMPEG 6. Below, is the command used:
ffmpeg -hwaccel cuda -f lavfi -i "amovie='$audio', asplit [a][out1]; [a] showspectrum="fscale=$escala:scale=sqrt:legend=enabled:stop=12000:win_func=flattop:mode=separate:slide=scroll:s=1638x952:color=rainbow:overlap=1:fps=60:gain=1" [out0]" -vf "drawtext=fontfile=\"/usr/share/fonts/truetype/noto/NotoMono-Regular.ttf\": text='%{pts\:gmtime\:0\:%M\\\\\:%S}\:%{frame_num}': start_number=1: x=0: y=0: fontcolor=white: fontsize=24,subtitles='$legenda':force_style='FontName=Minion Pro,Alignment=8,FontSize=24'" -b:v 16M -c:a ac3 -b:a 320k -c:v h264_nvenc "$saida"
Operating System: openSUSE Tumbleweed (kernel: 6.6.6-1-default)
GPU: RTX 3060 Mobile
It is possible to notice the frequency removal in some parts of the song. However, for the most part, the outcome is really good.
27 авг 2024