extracting multiple noises from a single track

I am afraid that i am a novice with audacity, but have used the simple features for many years.

I am doing some research and what i need to do may not be possible, but i figure this would be the best place to ask.

I have a broadband noise (not music or anything nice), that is a mixture of three separate broadband noises that are produced in close proximity to each other. My ultimate goal is to be able to be able to say with reasonable certainty which of the three sources is the most prominent without having to take a recording from the individual sites. The problem is that the three sources sound relatively similar to each other, and i cannot get a completely isolated recording from each source so i can get pure defining characteristics of that source. I figure that the individual sounds will be louder if i were to take a recording from each source. Sadly they are approximately 2-3cm from each other so there is significant overlap.

To add another layer of complexity, i will have 100 or so of these noises to divide into three. All of them are subtly different, and each would have different proportions of source sound in thier make up. I will therefore need this process to be automated!

I dont need to actually hear the isolated sounds with any clarity, but i do need a computer to be able to tell me vaguely what it believes are relative proportions of sound from each source. E.g. 10% from source 1, 80% from source 2, and 10% from source 3.

So can audacity be programmed to divide my noises up? I note that there is a lot of people who seem to be able to isolate / remove vocals or bass etc from music. Presumably this technique requires someone to code in the defining characteristics of these sounds and extract them from the track. Since my ‘noise’ is relatively constant with little in the way of big variations seen in music i am hoping this will prove a little easier?

I also accept that this is something audacity was not designed to do, and i am happy to pay a developer to create a custom build audacity version for me. If it cant be done however, then that will probably save me a lot of time and money.

I would be interested to hear your thoughts


I note that there is a lot of people who seem to be able to isolate / remove vocals or bass etc from music.

A lot of people asking about it and seem to be making it work. Taking the bass out is the digital equivalent of turning the tone controls on a stereo system. Nobody has to identify which bass violin did the work and separate it out from the cello sitting next to it playing similar notes. Everything goes.

Vocal removal is an arithmetic trick. If you subtract the left sound from the right sound in a very well done stereo show, many times, the singer directly in the middle will vanish. That’s it. The system has no idea what or who the singer is and anything in the middle like bass and drums goes too.


As a general rule, you can’t take apart a mixed show or song into individual instruments. I suspect you’re trying to take this to the next level of which sound is more objectionable than the others. That gets into how your ear works and why a crying baby on an airplane causes you to want to perform homicide…

If you have a stereo recording and the noise sources are in different parts of the stereo field then you may* be able to isolate individual sources using Extraboy pro

[ * if the sounds are being reflected by say walls then completely isolating one source by this method will not be possible ]