Comments

SP. wrote on 2/8/2024, 7:32 PM

@Dave-Wallin-Eddy The ad was likely talking about SpectraLayers which is included in the Suite version of Sound Forge Pro 17. The current SpectraLayers version uses the Demucs AI model, made by Facebook, to split songs into multiple stems.

But you don't need SpectraLayers. You can also download Ultimate Vocal Remover 5, which is a free tool. It not only includes the Demucs model, but also additional models which can get better results in case you want to only split vocals and music.

It's recommended to have an Nvidia GPU, since the demixing is much faster with it. Takes only a couple of minutes for a complete album. If you don't have such a GPU you can also run it on the CPU, which will take about 15-20 minutes per song.