I know - sounds a bit strange. As a matter of fact it DID sound a bit strange!

Our brass band had a concert recently when I was away on a business trip, so I handed my setup to a friend and he did a good job of setting up like I had told him to and recording the concert.

But I hadn't realized that they were going to try to have a combined performance of band and bellringers and the way they lined up the bellringers it turned out that they were strung out in a single line across the front of the venue, but quite a ways in back of the microphones. Had I been there with my kit I would have tried to mic the bells somehow, but no way he could do it.

As you might expect, the bells were almost inaudible - scanning through the recording later I actually missed the bell solo becase the amplitude trace looked about as flat as the brain wave of a carrot. In spite of which you could sort of hear - something.

I was able to boost and otherwise clean up the solo so it sounded - well about as good as a handbell solo could sound - to my ears sort of like milking time on the farm with the cows coming back to the barn. There was very little mic/mixer/recorder noise so even after a 20 - 30 db boost the noise level was pretty fair. There was some kind of heavy rythmic thumping, but low frequency roll off pretty well suppressed it.

Always enjoying a challenge, I then tried to figure out how to selectively boost the bellringers in the piece they did with the band without of course making the band any louder.

I'd love to hear any ideas others might have, but what I wound up doing was first to look at the piece in Izotope RX, from which I could see that the spectral plot for the bells was quite different from the spectral plot for the band. So I trained the nose reduction module on a solo section of the bells and then ran the denoiser, having selected the option to output the noise in place of the cleaned up audio.

As I had hoped (???) after playing around a bit this gave me a track with most of the bells but practically none of the band. I mixed the "noise" track back in with the original as a "reinforcement" and voila, was able to boost the bells (particularly the lower bells) so they held their own much better against the full band.

I'm not claiming that this was any work of genius, and I still don't care for the bells that much, but at least it made the bellringers happy. And kind of fun in a way.

I just hope and pray that they don't add a bagpiper the next time!

Now, that's just clever! I may have to borrow that idea some time. Love this forum, everybody contributes..... / Battle Vaughan/miamiherald.com video team

Thanks for the kind words. Not sure it would work to boost a trumpet in a brass quintet because you'd be dealing with the same sort of sound spectrum - but it might work with the bagpipes. I think the key would be having sufficient dis-similarity in the spectra you were trying to separate.

Something I sort of learned from all this was that bells are quite acoustically interesting. When you look at the ordinary sort of amplitude based waveform display, the bells appear to be low amplitude but sound relatively louder than they look. The way the human ear processes the signal has more to do with their perceived tone and loudness than the actual energy/frequency distribution. At least that's the way I remember it.

Of course this is true to some extent of all instruments in that people actually synthesize what they're hearing from the harmonic series rather than from the fundamental tone itself. There was an old organ builder's trick wherein they would pair up two shorter pipes with higher frequencies and carefully tune them so that instead of hearing two separate pipes, people heard the lower note that would have been generated by a much longer pipe that wouldn't have fit in a small church or would have been too expensive.

Just think of hearing the non-existent notes in a Barbershop Quartet.