Flute Tone Investigations
Perhaps one of the most difficult parameters of the flute to deal with scientifically is tone, but of course it's also one of the most important. And there are risks in just trying! The flute buying public is ever ready to condemn the relevance of science in this area - you'll probably remember the hostile reaction to John Coltman's clear demonstration that listeners could not discern the material from which a flute is made. But that shouldn't stop us from trying to get scientifically closer to the topic, although it will certainly encourage us to present what we find carefully!
I'll be the first to admit that I don't have the answers I seek yet, so this is going to be a voyage of discovery on which you are invited and may well be needed.
It's the player, stupid!
One of the criticisms of scientific investigations has been the inevitability of needing players. (A flute won't play itself, and any suggestion about use of a mechanical blower seems to attract even more criticism!). How is the scientist to dissociate tonal effects introduced by the flute from those introduced by the player? A recent paper from Gregor Widholm employs an interesting approach, as well as some interesting findings. You can find the paper at:
and I'd certainly recommend you read it.
But let me draw your attention to the statistical approach he used. He was examining the relevance of materials in metal Boehm flutes and obtained seven from the Muramatsu range, differing from each other only in that they were made of 7 different alloys, ranging from silver-plated base metal through the various gold alloys to full platinum. (If you look at the photo of flutes in the article, you'll see they all have B feet but one, but I don't see any signs of this affecting the outcome.)
He also drew on seven players, all orchestral professionals. Needless to say, they all had their different tones, as amply illustrated in this cepstrum, where all seven players play just one of the flutes....
By then averaging the responses each player got from each flute, Gregor was able to demonstrate that all the flutes gave essentially the same response:
According to the paper, the differences in the seven tightly-packed curves above amount to less than 0.5dB. Good to remind you that a 1dB difference is the smallest change a trained ear can detect in a fixed tone in ideal listening conditions. A 3dB change is the smallest an average listener can detect in normal program material, and 10dB (or 1 Bel) is the difference that listeners perceive as a doubling in loudness. So 0.5dB is certainly not likely to be perceived by listeners, even professional listeners.
Now the similarity in tonal output should come as no surprise to us. We know that all the flutes were made of metal by a top manufacturer so no issues of smoothness, stiffness or porosity likely to surface there. Because they were made by the same maker, using modern controlled manufacturing processes, they're pretty likely to have precisely the same dimensions. Certainly, if one had stood out in the pack above, it would have been easy to see, and imperative to look for reasons why. But it didn't.
I should mention that Gregor wasn't relying on just these spectral images - he also carried out controlled listening tests that reached the same conclusions. Indeed, you can hear some of these for yourself if you go to the site mentioned. He has a sound grab of the seven players playing a note on one flute - and getting very obviously different tones from it - and one player playing the same note on each of the flutes, and I wouldn't even try to tell them apart!
Gregor wraps it up hoping that his experiment will put paid once and for all to the suggestion that different metals produce different sounds. Fat chance. I'd be convinced, but I don't think the flute buying world will be. They will go on believing that upgrading from silver to gold, or gold to platinum is going to completely revolutionise their playing, and the manufacturers are certainly not going to tell them otherwise. And once they've stumped up for the tenfold increase in price they are certainly unlikely to go back. Dismal news for the science of acoustics, but probably better for the economy this way!
It was perhaps a shame that Gregor didn't go on to do one more test - which flute did the players prefer? Blindfold the players and let them play the seven flutes for as long each as they liked. Have someone also blindfolded hand them the flutes, fed to them in a pseudo-random order.
(Heh heh, all that clattering noise is the sound of two blindfolded people trying to pass expensive metal flutes back and forth in a pseudo-random way.)
It's possible that the players themselves could discern differences in performing qualities that might lead them to prefer playing (not just listening to) some of the flutes over others. It would be fascinating to see if there was consensus, and to what the differences can be attributed. But maybe that's a whole new experiment for another day.
What's left to do?
So where do Gregor's findings leave us? Isn't the work all done?
I think not. Firstly, unlike Gregor, we're not concerned with dimensionally-identical flutes. In the far more diverse world of Irish music, we find the occasional Boehm flute, the even less common Boehm bored flute with simple-system fingering, vast numbers of Rudall style conical flutes, heaps of Pratten's style big-bore conical flutes, an increasing number of small bore conical flutes like my Grey Larsen Preferred, and some plain cylindrical instruments thrown in. We know from our own experience that they sound different. (Well, we think we know!) Now we'd like to know what those differences comprise.
And within those broad categories, we have the work of different makers, with more subtle variations of dimensions and features - based on different originals, different interpretations of the same originals, with differing levels of precision, different embouchure holes, free-flowing D feet compared to tapering C feet, keyless, partly keyed, fully keyed and so on.
And our materials are not all "perfect containers for air" - smooth, stiff and non-porous. We have woods ranging from the medium weight (and relatively porous) furniture timbers through to the densest and finest timbers on earth. We have rough and smooth bores, oiled, waxed and lacquered. We have PVC and Acetyl, and metals in the form of headliners. What are the differences between lined and unlined heads? How "imperfect" does a material have to be before it starts screwing things up? And can a material be "too perfect"? Should we just toss African Blackwood for Black Acetyl?
And we also have a much more diverse community of players. Gregor's professional players will probably have all come through a conservatoire system and have a lot more in common than players in our community, who range from self-taught in isolation to well-tutored in flute-players' paradise. And even some who have come through the conservatoire system before they came over to the dark side.
And while Gregor needed to isolate the effect of the players, to concentrate on the effects of the metals, we need to isolate it sometimes (to determine the effects of the flutes) and then reincorporate it (to see how different players can drive the same flutes in different directions). What is the range of tone we can expect from one flute? What are the indicators of "desirable" and "undesirable" tone traits? What playing approach will maximise the desirable and minimise the undesirable? Can changes to flute design make this easier? And, perhaps relinquishing acoustics for the better-paid realms of psychology, is easy good, or is manageable pain better?
So, plenty of work to be done! I'll tease out these questions and add pages as time allows. I suspect the first thing we need to do is test whether Gregor's "cepstrum" approach - taking the spectrum of the flute as a whole rather than a series of spectra of individual notes - is entirely applicable to our flutes. I suspect we might find it isn't, and for an interesting reason.
I'll leave the last words of our introduction to the series to words attributed to Lord Rayleigh (John Strutt, Cavendish Professor at Cambridge, and author of the two volume Theory of Sound, published in 1887-1888, about when Rockstro was writing The Flute):
" What we cannot measure, we do not know".
Thanks to John Coltman and Gregor Widholm for lighting the way!
The Flute Tone Investigation Series so far:
Created 11 October 2009