Showdown - conical / cylindrical

[Gordon]

Most of us guessed right, it seems. Still, I seriously doubt any of us, not told beforehand that a cylindrical was one of the choices, would have said they weren't both conical. What's interesting (and vaguely amusing) is that a large portion of us preferred the sound of the cylindrical.

[Calum Stewart]

To refer back to what I was trying to prove:

.... " to prove that bore type (between a wooden pratten type model, and wooden cylindrical model) makes no difference whatsoever to the sound of a flute."

So, I happily accept that the results prove more people can tell the difference between the two bores systems! Although the majority of people stated this being very hard to choose, the interesting thing for me in this test, are the different remarks to each flute - some quite contradictory!

So to conclude: More people (although not a lot in it) seemed to prefer the cylindrical flute, despite obvious previous leanings toward conical flutes. Although making a strong statement at the beginning towards aiming to prove my goal, I was fully prepared to be proven wrong, and I actually found this test exciting!

Thanks everyone!

[david_h]

I wonder if you were tending to playing each flute in a way that suited its characteristics rather than for what you might want from it when performing. I thought Flute A sounded "more like you" on Earlywood but for recordings of a steady play through of tunes in your kitchen I preferred the less 'dramatic' sound of Flute B. Were you pushing the conical flute harder on Earlywood ?

I look forward to hearing more of whichever you prefer.

(I guessed right)

[Nanohedron]

I am still nagged by some doubts, though (I guessed correctly, BTW): How much of the difference in tonal color might be attributed to embouchure cut, and fingerhole size?

[I.D.10-t]

Well I guessed wrong. Kind of figured I did when I went back and re-read the description of the Rudall flute in the original opening post.

[Akiba]

I guessed correctly, with an emphasis on the word "guessed". I, too, would not have known that one was cylindrical; would have thought both were conical. I did like B better because I thought it was smoother, more balanced, with a stronger low D. Perhaps that's the key (and I find it similar when I try to play my Boehm as an Irish flute): the cylinder needs to pushed harder to get "the sound", whereas for the conical "the sound" is more a part of its natural character.

[jemtheflute]

I too "guessed" correctly, and preferred A. Taking the points I've discussed above into consideration, my decision was ultimately based not on preconceptions of what to expect sonically from each type of bore but on knowing that Calum thinks he "finds" more in/from the cylinder flute than he does in/from conoid ones, and on balance I think the clip (possibly excepting in Shores of Lough Gowna) overall shows him to seem to be "finding" more in/from flute A than he does in/from flute B. But that query-exception for the 3rd tune points up how hard this is - to set up fairly as well as to judge.

Unless I've missed reading it, I don't think you have told us what flute B actually is, Calum. Come on, you can spill that now.

[Bill Wolfe]

Interesting! I had a pretty strong impression of which was which, partly informed by hearing Calum's demo of the Peter Worrell cylinder flute but also because A sounds more "Pratteny" and B more "Rudally" to me, if that makes any kind of sense. It's like the spectrum was shifted over to one side but the basic tonal gradient still held. Sorry, I'm on my second glass of Chardonnay. Anyway, I "preferred" flute B in an abstract sense, meaning I lean towards a tighter, more nasally sound as an ideal (words really aren't up to the job of defining flute tone). But I thought the performances on flute A were generally more exciting and engaged. Of course, everything played was at a very high level of musicianship. It's just that Flute A seemed to lead the player into more interesting and unexpected places. The one performance that really stood out for me as highlighting the virtues of Flute B was the Banks of Lough Gowna. Like Jem says...
BW

[Nanohedron]

...A sounds more "Pratteny" and B more "Rudally" to me, if that makes any kind of sense.

I can see why you'd say that. Breadth as compared to, umm...restraint, maybe, which is a funny thing to say about a Prattenesque design, isn't it. I chose B as the conical because tonewise it struck me as more intimately familiar somehow, and so in a trad world more dominated by conicals, I figured that must be right. While ultimately I liked each on its own merits, I had a preference for B for myself, but then I've gone Rudall-style and never really looked back, and the comparative difference may account for that. It was a hard choice, though; the guessing was easier.

[Terry McGee]

Now, as you all know, there isn't much that can't be confused further by applying a load of measurement software, and this is no exception. I took some samples of Calum's recordings and ran them through Fast Fourier Transform to see if spectral analysis would reveal all. I had little expectation, as, even if I could detect a difference, what would tell me which was which? More of that later.

I picked the middle tune and compared cepstra of both of them. A Cepstrum is simply a spectrum, not of one note, but a collection of notes. You can think of it like an audiogram used by forensic scientists to provide a "voice print" of a suspect. There were small differences, but nothing I'd want to rely on in court.

So then I compared individual spectra of the ending notes of all six pieces. Again, there were a few differences, but nothing earth shattering. But how to interpret these differences? I thought to play the same notes myself, on an original Pratten's Perfected, and an original cylindrical bore 6-hole, 8-keyed flute. But all I proved was that Calum's recordings were played by Calum, and mine were played by me. My recordings sounded different, and measured different. In other words, I proved what we already knew, that the difference between players is bigger than the difference between flutes, providing the flutes aren't ridiculously different.

Which perhaps brings us to the question, why are these very different flutes not that different in tone? Jem has identified the issue - that the hole size is responsible. The venting of a flute is broadly proportional to the diameter of the bore divided by the diameter of the first open tone hole (but obviously modified by the next one or two holes). The Boehm bore flute starts off life with a bigger bore, but the fingerhole sizes are about the same, so it should end up about as dark if not (at least in theory) a bit darker.

But Calum's two recordings do sound quite different, but don't measure all that different. What's going on there? I suspect (listening to them) that the difference isn't so much spectral, but noise and other non harmonic artefacts. The second flute is much cleaner, which some people prefer and some are less keen on.

I'm not sure that technology at this stage could easily identify and quantify that. But, more importantly, is this a factor of the two flutes' bore designs, or a factor of their differing ages and condition?

Terry

[jemtheflute]

is this a factor of the two flutes' bore designs, or a factor of their differing ages and condition?

Indeed - Calum could have used his new Worrell copy of the RC if flute B is modern to address that, but I also suspect embouchure cut is significant. I assume he used his modern Bohm style Mancke head on the RC. Whatever flute B is, an old Pratten or (as I suspect) a modern one, it probably has an elliptical embouchure whereas the Mancke head presumably has a rounded rectangle, and who knows what kinds of undercuts are involved. Unfortunately one cannot use the same head between the two types of flute to try to eliminate that variable, though one could have CNC machined, untweaked cuts made in a head of each type....... Whether or not the heads are fully, part or unlined could (allegedly) have an influence too, ditto any embouchure bushing or edging. FWIW, I did think Calum sounded a bit "airier" or even "hissier" on A than on B.

[LorenzoFlute]

I guessed right, but prefered the A sound. I have the feeling that with a Rudall type of flute with a sharp edge on the emb hole it would have been more difficult to guess though. That Pratten style flute used in the clip had a very distinctive kind of voice which is not necessarily representative of the "conical sound".
Can you tell us the make of the flute Calum? My guess is that it was an Olwell Pratten (believe it or not, I'm not a huge fan of these...)

I picked the middle tune and compared cepstra of both of them. A Cepstrum is simply a spectrum, not of one note, but a collection of notes. You can think of it like an audiogram used by forensic scientists to provide a "voice print" of a suspect. There were small differences, but nothing I'd want to rely on in court.

So then I compared individual spectra of the ending notes of all six pieces. Again, there were a few differences, but nothing earth shattering.

[...]

In other words, I proved what we already knew, that the difference between players is bigger than the difference between flutes, providing the flutes aren't ridiculously different.

See, as I said before to my ear the differences were quite huge. Not so much to the computer apparently. Now, can we really trust this scientific measurements? Looks like they're not telling the full story... Off topic though.

[Terry McGee]

See, as I said before to my ear the differences were quite huge. Not so much to the computer apparently. Now, can we really trust this scientific measurements? Looks like they're not telling the full story... Off topic though.

I agree the difference between the sounds of the two instruments is quite evident. But that doesn't mean the differences are harmonic in nature, which is what I was testing for, and is what we naturally assume a difference in tone is due to. As I mentioned, I think the most obvious difference is noise. It should be possible to measure the "signal-to-noise ratio" (to borrow the Hi-Fi term), but it's not straightforward. It's easy in Hi-Fi, you measure the test signal with the noise, then turn off the test signal to measure the noise only. In the case of the flute, stopping blowing also turns off the noise!

Comparing spectra of the Low D notes of the two flutes, Flute A noise appears at least 6dB higher than Flute B's, when compared to the highest harmonic. But really, we'd need to compare the averaged noise to the summed energy in all the harmonics, which takes a bit of computation.

The "shape" of the noise floor is also different between the two. Flute A has a flatter wide-band noise floor, while the noise in Flute B is more centred in the 1 to 3KHz region. In theory this should make it more noticeable, but it also puts it among the strong harmonics, where masking will make it less noticeable. (We notice noise less if there is a strong wanted signal in the same octave, and there are 5 strong harmonics in that band.) So it wouldn't be enough to compute the noise, we would also need to apply weightings to adjust for known psychoacoustic factors.

So, I wouldn't write off the science on the basis that I measured the obvious parameters, and it turned out not to be them.
That discovery alone is of interest.

I'd be interested to hear from Calum what the approximate miking distance was. Was it close enough to give us the player's perception, or far enough away to give us the listeners' perception? We might well find that they would measure differently.

It would be also of interest if Calum had an explanation for the differing noise levels and characters.

Terry

[LorenzoFlute]

I wouldn't write off the science on the basis that I measured the obvious parameters, and it turned out not to be them.
That discovery alone is of interest.

Agreed. I only don't like when scientists say "your perception is wrong because of what we measured", when they only measured some and not all the constituents of something (because the others are "more difficult", or unknown)

[csoroz]

I'd be interested to hear from Calum what the approximate miking distance was. Was it close enough to give us the player's perception, or far enough away to give us the listeners' perception?

Here Calum details mic position: viewtopic.php?p=1089577#p1089577