If you were trying to strip the VOCALS (i.e. get a karaoke version), you could turn to Blind Source Separation. Stay with me here, because I'll get to background music in a minute. BSS is a way of separating mixed audio signals. There are a handful of BSS methods, but here's a VERY SIMPLIFIED explanation of one method:
A studio often mixes a song so the vocals are positioned in the exact "center" of the recording. The instruments are placed slightly off-center. Remember, with a stereo recording, we have a left and right channel but NO "center" channel. To put something in the so-called center, you just add the same amount to both left and right channels. This means that vocals in the "center" are on both the left and right channels at the same EQ and level. By reversing the phase of the left channel and adding it to the right, you cancel out the center portion off the song.
This is a common method that can produce pretty good results, but it's not perfect. Oftentimes you still hear a slight echoey voice in the song.
Another method is to play with the frequencies. You can try to cancel out the pitch range where the vocals occur the most.
But you want the opposite: louder background music and quieter vocals. This would theoretically try to keep the center and remove the rest of the stereo image. Unfortunately, this is much harder to accomplish, but here are some methods to try:
(I'm assuming you're on Windows, if not, let me know.)
Method A
Use
Transcribe. Load your song, combine it to mono, then use a low pass filter and high pass filter to remove sound outside of the singer's vocal range.
Method B
Get a free/shareware VST host application such as:
-
MiniHost
-
Audacity
-
Buzz COMBINED with
Polac VST Loader
Load the
Kn0ck0ut plugin into your host application. Apply it to your song.