ChatGPT Flute Making Advice

[shrines]

Thanks, Geoffrey. The easiest thing for me to do is modify my bore taper and see what happens. I'll give that a try before I dive back into all the programming necessary to recreate Yang's work.

[Terry McGee]

What originally struck me as interesting about the ChatGPT response was the statement that a longer taper flattens the higher notes. I guess the only way I'll find out is to extend the taper and see if it makes a difference. Now I have to figure out whether to extend the taper toward the headjoint or toward the foot.

Isn't the Reductio ad Absurdum of a "longer taper" the cylinder? And we know that a cylindrical flute flattens the upper notes.

So does a tapered tube sharpen the upper notes, or flatten the lower notes to equal them? "Hello, ChatGPT, I have a further question for you....."

[Tunborough]

Bore Diameter: The bore diameter of a flute affects the tuning of the high notes, including those in the third octave. A larger bore diameter will produce sharper high notes, while a smaller bore diameter will produce flatter high notes.

Interesting. This conflicts with my own experience. Using a larger bore for a given key most certainly does NOT sharpen the high notes--quite the opposite.

That's the thing about chat-bots. When they don't know the answer (and really, they never "know" any answer), they make something up out of something they've read somewhere. It sounds plausible, and is expressed with great confidence, but it may be absolute balderdash.

[Geoffrey Ellis]

That's the thing about chat-bots. When they don't know the answer (and really, they never "know" any answer), they make something up out of something they've read somewhere. It sounds plausible, and is expressed with great confidence, but it may be absolute balderdash.

Never trust a chat-bot. And furthermore, what the hell is a chat-bot?? I've never heard of one before reading this thread. But based upon what I've heard so far, they are not inspiring a lot of confidence.

[paddler]

That's the thing about chat-bots. When they don't know the answer (and really, they never "know" any answer), they make something up out of something they've read somewhere. It sounds plausible, and is expressed with great confidence, but it may be absolute balderdash.

Never trust a chat-bot. And furthermore, what the hell is a chat-bot?? I've never heard of one before reading this thread. But based upon what I've heard so far, they are not inspiring a lot of confidence.

Indeed! But how do we know you are not a chat-bot, Geoffrey?

Alan Turing wrote a nice paper about this problem almost 75 years ago. For a quick summary, read about The Turing Test.
We are finally approaching the point where the behavior of machines and humans is getting very difficult to distinguish, and in part for the very reason that Tunborough stated!
I enjoyed Tunborough's summary, in part because it is true, and in part because it is a very good description of the behavior of some people I have encountered.

[david_h]

We are finally approaching the point where the behavior of machines and humans is getting very difficult to distinguish, and in part for the very reason that Tunborough stated!

Depends on the context. We can ask the posters in this discussion to explain what they say, why they think that way and give references to other sources of information. If a chat-bot would do that it would be more convincing - and more useful.

On the topic. My Burns flute is also harder to play in the third octave that my antique Metzler. But the antique with the Burns head on it is not that different to how it is with its own head. So maybe the difference is in the body rather than the blow-hole cut.

[Tunborough]

I enjoyed Tunborough's summary, in part because it is true, and in part because it is a very good description of the behavior of some people I have encountered.

Ah, yes, that crossed my mind when I was writing that. We are all capable of it. NPR did an episode of This American Life a couple of decades ago on the dangers of having a not quite enough knowledge. https://www.thisamericanlife.org/293/a- ... -knowledge

[waltsweet]

Steve,

I'm in the throes of designing my new Bb drumcorps fife, EMMETT model. The tuning issues are slowly resolving.

Simplified rule: The more taper (greater taper rate, more inch-per-inch), the greater stretch of the octaves. I avoid saying "the top octave is flat." Boehm's curvilinear* ("parabol-ish") head worked by lowering the octaves progressively. Of course, this mod did the job. Analogy: when I'm setting-up my lathe to cut a tapered profile, I must first set the relationship of beginning-to-end (the overall taper), then I adjust the basic diameter. So in acoustic design, it's important to address the overall relationships first (and to rephrase the problem). The rule about increased taper rate is supported by Rayleigh's Rules.

Taper rate (in the body) is one way to adjust the octaves. Another is the endstopper position: moving closer to the blowhole will stretch the octaves progressively. It seems these parameters comprise two ways to do the same job, but it's not quite that simple. I have found that the endstopper position has a great influence on the ease of producing the notes in the third octave. These high notes need to be available under normal circumstances as enhancements to the lowest notes. The low notes need the highs as upper partials; without them, the lows lack complex character, the flute lacks tonal flexibility. I have played flutes that can't get the basic high notes (G3, A3), and such flutes are lacking in playability (even when the low notes pass muster on the tuning machine). By making incremental adjustments, a serviceable position can be found for the endstopper.

The notes G3 and A3 are basic "anchor" notes. Sure, you want a flute that can play all the notes in between, but for Gestalt, I think we need to find endstopper position, length and rate of taper 12-24, length and rate of taper 0-12, in that protocol. From there, you can look for complex fingerings in the third octave. About the 12-24 bore: making it smaller will raise the third octave slightly, lower the second somewhat, and raise the first octave quite a bit (supported by Rayleigh's Rules).

You mentioned the fingerings based on the second (fourth) and third harmonics. The principle is easily seen on the fingerings for Boehmflutes. For example, to play F#3, we open the F#-tonehole (to sound its 4th harmonic = F#3) and at the same time, open the B-tonehole (to sound its 3rd harmonic, also = F#3). However, we have to be careful on the diatonic flute because most holes take a step of two semitones. To do so, we make the holes smaller and higher up the bore from Boehm's Acoustically Correct Position; the B-tonehole especially will pull the pitch too high. The problem is verified on the drumcorps fife: To play F#3, we open hole #2 (B); the output is 40 cents sharp because the B-tonehole is too high. To play G3, we open hole #2 as before; but in this case, output pitch is OK because the proportions have changed and hole #2 is not too high up the bore now. I don't know if your chat-bot has been programmed to think in these terms.

*In Die Flöte und das Flötenspiel, Boehm only said it resembled a parabola, but so do many curves. He never said it was mathematically a parabola (a paraboloid); the math nerds must have hijacked this notion. My next trip to the soap-box will discuss Boyle's Law -itis.

Walt Sweet

[Terry McGee]

Agreed, Walt, Boehm only said resembled. It was Richard Carte who went totally overboard on the topic in a sales puff-piece promoting his flutes. My favourite:

"The parabola-head-joint seems to effect that for propagating sound, which the parabolic reflector does for propagating light. The vibrations are concentrated in, and propelled from the one, as the rays of light are concentrated in, and transmitted from the other, both with superior velocity and power."

So the mighty parabola not only makes lights brighter and sounds louder, but faster as well. And I thought the Speed of Light and the Speed of Sound were pretty much regarded as constants, well before Boehm and Carte!

[buha]

My model is is quite good (I am within 10 cents on all notes in the first two octaves) but the third octave notes are 30-60 cents sharp. I have been hunting the Internet for definitive guidance for a month or so with no luck. So yesterday, in act of frustration, I succumbed to all the hype on the Internet and main stream media I submitted a query to ChatGPT.

May I ask if your model is open source Steve?

3 yrs ago I worked on a simple model that I could use to figure out where to drill finger holes for making kavals. It makes many assumptions, and it's quite stupid in places but I did some homework, incorporated some empirical formulas from acoustics research and I was able to use it to build a Kaval in D.

https://github.com/buha/kaval/

Like you , I soon realized it gets tricky in the upper registers and found out about acoustic impedance and Paul Dickens's thesis. I saw that the thesis has code layed out in the PDF and copy pasted every page, wrote a Makefile for it and some scripts to visualize the data.

https://github.com/buha/flute-thesis/

I didn't quite get to do much with that due to shifting priorities but now I am back at it and I'd be willing to build am impedance head, sweep some frequencies across my kavals and do some math to compute the impedance and compare it with model computed impedance, see how far off the model is.

So hence my question, do you have anything open source out there ?

[waltsweet]

Do you own/have access to a flute with tapered bore but Boehm fingering? I have a Badger here (c.1862). It has an interesting, complex tone, but certain other issues. Claire Soubeyran (RIP) used to make flutes like it. It's always insightful to start with something that's already working, and seeing how far you can push it in a new direction. Some of the problems can be isolated that way, or put in better focus for further consideration.
I've never played Boehm's 1832 model (old-style keywork); there may be something to be learned there.

1. Endstopper position and the availability of higher partials: still important.
2. Another schtick of mine is this: in our beloved design of the wooden flute, the side-holes (toneholes) add volume to the bore (Boehm avoided this, mostly). As a result, the part of the flute with the tonehole lattice is effectively larger in diameter. Benade said it was "lumpy;" I say it's fatter. These complications will definitely rear their ugly head in the third octave. I proved this factor to myself when I adapted a 6-hole fife to a 10-hole design. The solution was to ream the fife smaller in the the lattice zone. So now it appears that the taper in this zone should be treated separately, not as a continuum of taper from there to the head. There's more to this than just the rate of taper below the head, more than just making a new reamer.
3. Recent efforts have brought my attention to the zone between the top tonehole (C#) and the head. It is analogous to the zone between the 12th and 24th fret on a guitar. On the flute's first octave, a pressure antinode lands here. But there's more to it: when I reduced the diameter here, the 3rd octave went up slightly, the 2nd octave went down moderately, and the 1st octave went up significantly. After that, I reduced the bore in the tonehole lattice, making a better instrument overall. I had found new control over the 3rd octave (in relation to the other two). Refer to my point #1. Repeat as necessary!
Walt Sweet

[paddler]

Interesting post Walt! A lot of the issues you mention are examples of chambering in the bore. The cavities left by closed/keyed tone holes, the cavities left by closed fingered tone holes, the cavity left by an extended tuning slide, etc, are all examples of chambers in the bore. They all have an effect on tuning. Rayleigh's Rules explain the general effects of chambers and other variations in the rate of taper of a bore on the tuning of notes whose nodes and anti-nodes fall in their vicinity.

A study of antique instruments and flute making literature makes it clear that, historically, the best makers have been well aware of these effects for a few hundred years and have utilized both chambering and varying rates of taper in the bore to try to influence both tuning and ergonomics of their flutes, as well as to compensate for the effects you mention. It is a complicated process though, because the effects of any modification can't easily be isolated, so you end up being forced to do a long sequence of experiments and successive incremental changes and measurements.

Even producing a bore map that shows the location of the nodes and anti-nodes for all the notes in the relevant acoustic range for the flute is non-trivial when the bore is non-uniform. I think the generation of such a map, even if approximate, would be one of the more interesting and useful outputs from a computer modeling project, by the way.

Anyway, all of this explains why the best flutes don't tend to use a uniform bore taper, and instead incorporate varying rates of taper, sometimes including localized expansions of the bore at section ends, through the use of a set of reamers (or a non-uniform reamer) and localized back reaming. I think it also explains why the vast majority of makers choose to build directly on the work of others, by starting out copying a bore profile and tone hole lattice, rather than designing from scratch. Some of the flutes we enjoy playing took decades or generations to arrive at their complex tonal, ergonomic and tuning balance that we find so pleasing to play and listen to. As you say, it is insightful to start with something that is already working and see if you can slightly improve upon it ... and at what cost.

[waltsweet]

Thanks, Paddler. I've identified this upper quarter as a critical zone, not getting the attention it deserves, finding it unfairly grouped with the rest of the tapered section (i.e., the lattice). I used a flute fish to find the physical limits and I moved the tuning gap accordion-ly (as we say in the music business). My modern, Boehm-system P _ _ _ _ piccolo has some serious tuning issues that I improved with an insert in the zone. Similarly, I'm getting the strong idea that bulges in the bore (as those at the beginning of the RH section) are really detrimental. My Shannon used toneholes that were undercut above, and while this feature corrected the pitch, it didn't address other important issues.

There's another challenge with the diatonic flute. Of course this term means that the flute's design is centered around a diatonic scale (of whole steps and half steps), but each sharp (a note sounded when a NC key is opened) has different acoustics than its neighbors. When sounding G# for example, the flute has three cooperating toneholes: G#, G and F# are all open, the main TH plus two semitones immediately below. This effect aligns with Boehm's principle of open standing keys. G# can have a stretched octave, so the bore should have less taper here (as reflected in the bulge described above). In contrast, when we go from A# to B, we close the A# (Bb) key. Now B does not have the support, and it needs the full taper. We're trying to get the bore to support two different acoustical modes.

I'm getting good results by making changes that affect groups of notes, rather than discrete chambering.

Walt Sweet