Is there any way to subtract or cancel background music picked up from a microphone, if I have the isolated background music?

Here's the situation: I'm editing a lecture presentation. This lecturer has spiced up his presentation with background music and sound effects, which worked well during the live presentation, but needs to be removed for the digitally published version. There are many sections where he is speaking while the music is playing, so that the music cannot simply be cut out.

I have two audio tracks: the first is the recording from the speaker's microphone which has picked up the background music as well (the background music is quiet, but definitely still audible), I also have the recording directly from the speaker's powerpoint which is the isolated music and sound effects, these two tracks are time-aligned. My intuition tells me I might be able to use the isolated sfx track to cancel or subtract the background noise from the speaker's voice track. Is this possible?

I'm editing using the adobe creative cloud, specifically Audition for audio. Thanks in advance

• The simple time domain theory would be to just invert the phase of one track against the other… except for the fact that the spacial domain is not going to be quite so friendly. The room's reverb will still be audible, even after you have lined up the original track to the delayed audio received at the live mic. Once you hit that, then something like Isotope RX might start to come in useful, but it's not something I've ever had to test. Nov 14, 2015 at 10:29

My short answer would be that it is virtually impossible but depends on how good results you expect.

The solution your intuition draws you to, and which would work in certain cases, is phase cancellation. Basically, if you have two sound sources, exactly the same, and reverse the polarity of one they will cancel each other out. The simplest example of this would be a DC source at, say 5 volts sent to a speaker cone. This would push the cone outward. If you add to this another DC at -5V the sum will of course be zero and the cone will go back to its original position. With a complex waveform the same thing will happen.

The key here is "exactly the same". Any difference in volume, frequency response and phase will make more or less or all sound come through.

This approach would be possible with at least useful result if the speaker's microphone was in a fixed position and the soundtrack was played back all the way through with the speaker in position but silent. Ideally all factors should be as close as possible to the actual lecture. You could then exactly match the two takes in time and possibly, if done correctly, at lower the volume of the soundtrack (but gain 6db more noise).

In your case this wouldn't be possible. The original sound will differ too much from the sound recorded by the microphone from the speakers, each step introducing differences in frequency response and phase.

I think in your case you will have to yet and use EQ to as far as possible isolate the speech. The results would be far from perfect, especially since the frequencies prominent in speech is usually also prominent in music, but might be better than nothing. Some linear phase EQ:s are able to show the spectrum and make changes over time. This means a lot of careful manual work but might be worth it in the end. Aggressive editing in this regard will create artifacts in the remaining sound so there's always a trade-off.

There are of course programs, usually rather expensive, that deals specifically with removing unwanted sound from recordings and these technologies are getting more and more advanced. It was ten years ago I last used such utilities but back then Izotope had some impressive solutions to deal with noise. You might have some luck there. This seems to be their on latest tool set : https://www.izotope.com/en/products/audio-repair/rx/

Good luck! Although this might not be the answer you were hoping for I hope you find a working approach.

• I guess phase cancellation could be a good start to clean the sound. Nov 15, 2015 at 22:15
• @jonhatan smith: it could be if you had a recording of the sounds as they go through the PA to the microphone, in the exact same position and with the lecturer present. The original sound track will unfortunately be too different for phase cancellation to be viable. However close you'd try to match them you'd just end up adding more of it instead. Nov 22, 2015 at 16:39

i know this is an ancient thread, but it still shows up in search results! Pretty silly for the top reply to be "nope sorry can't be done".

Depending on where the speaker's microphone was, adjust balance (prevalence) of left+right sides of powerpoint's direct audio, then downmix to mono (assuming speaker's mic was mono). You may then be able to use the "Debleed" feature within iZotope's RX...8? or 9? or whatever. to reduce this.

The debleed is meant for when mic1 is recording Source1 but picks up a little of Source2, whilst you also have the recording from mic2 which was directly recording source2 - it will try to get rid of source2 from the recording of source1 - has worked in the past for me to various extents. I know this is too old to be useful but if someone new who newly developed a similar issue reads this... at least it will help someone.

I hope it's obvious enough what mic1, mic2, source1, source2, are supposed to be in this scenario - my explanation is a simple but solid one, hopefully.

• +1. It must be noted that RX has come on in leaps & bounds in the past 7 years since the question was asked. Mar 11 at 19:07

Adobe Audition has a built in Noise Reduction effect that works really well. Effects>Noise Reduction/Restoration. You can use this to capture the noise profile of the background music or other sounds you want to remove. Just find the longest section where only the unwanted sound is playing in isolation, and grab this. The longer the selection, the more precisely the Reduction can work. Then, select a section of the audio you would like to test, and open the NR settings to adjust the amount.

For full control, you will need to use the "Advanced" drop down, where you can choose a larger FFT size for greater Noise print fidelity. I suggest playing around with the other settings to find the best balance, because extreme settings will drop unwanted artifacts. I can elaborate on the basics of most of these settings, but I generally have had to just play around until I get the ideal result.

The good news in the case of vocal material is that you can generally get away with a lot of NR before losing quality.

This is a classic application of adaptive echo cancellation (speaker phones need it to remove the speaker output from the microphone input which would otherwise appear as an echo at the remote speaker's phone), usually done with the LMS algorithm.

No but you might improve it a little. Next time get a better recording of the speaker. And if they want the audio tell them to stop with the extraneous noise fx