Removing Bass Guitar

I’m not sure if there is much of an answer for my question but nothing ventured, nothing gained as they say.

I’d like to remove as much of the bass guitar as possible from an MP3 audio file so I can record and mix my bass to replace the original bass line.

Thru experimentation I have found that using the High Pass Filter at 300Hz and a roll off of 48dB/octave removes enough of the bass guitar to be acceptable for my needs. The problem is that after doing this filtering the result sort of sounds like what you might hear from a pair of earbuds placed on a table across the room or a 1960s vintage transistor radio. It’s not the lack of volume that’s the problem but a lack of overall fullness. I think what is happening is there is enough frequency content below 300 Hz in acoustic guitar, keyboard and other instruments that contribute to the overall fullness of the song. Even though these frequencies are not coming from the bass guitar they are also removed in the filtering. Even with my bass included into the mix these instruments sound thin and tinny. My bassline is adequately full sounding in the mixed recording.

So is there an effect that I can apply to the audio waveform either after the high-pass filtering or after my bass has been mixed in that would restore some of the overall fullness of the acoustic guitar, keyboard and other instruments to make them sound less thin?

One idea on how to maybe do this would be to identify each instrument in the mix (or even 2 or 3 of them) and add frequency content that is 1 octave below their lowest frequency. This could restore some of the lost fullness without adding back any bassline. If Audacity can’t do this is this a reasonable feature for a future version?

Any thoughts would be appreciated.

Yes, that’s the problem…

There is a lot of overlap with different instruments. All instruments (all natural real-world sounds) contain harmonics & overtones. A bass guitar only goes one octave lower than a standard guitar, and above that lowest octave the same notes can be played on a regular guitar, (A regular guitar also goes more than one octave higher because there are more strings.)

You missed one. MP3 gets its small, convenient files by re-arranging musical tones and leaving some of them out. So you don’t have pure, clean instruments right at the starting gate.

Audacity can’t be used to split a mixed performance into individual voices, instruments, or sounds.

Back up the individual sounds in your production before you mix down into the final sound file. Once you mix the show and close Audacity, that’s the end of the world.


Thanks for the replies.

Apparently technology exists to do what I am asking. Does anyone have experience with DeepRemix?

Audacity is mentioned a couple of times beginning around 13:00

Is this something that could be added to Audacity in some future release?

There is currently ongoing development work to integrate Tensorflow into Audacity. This is a big project, so it could be quite a while before it becomes part of a release.

Thank you Steve, I know you guys can do a great job :smiley: