"WhistleMark"-absolute measure of whistle quality

[eskin]

Hi y'all,

Since my company is in the audio DSP industry, I figured it was time to move beyond all the non-scientific discussion of cheap vs. expensive whistle quality and instead come up with a absolute, repeatable measure of whistle quality using a acoustic model of a simple 6 hole whistle combined with advanced harmonic modeling concepts. These same ideas have been used for years by Yamaha for their acoustically modeled electronic wind instruments, in our case, we essentially reverse the process.

The end result is a numeric value that we call a "WhistleMark" that can be used as an absolute comparison of whistle quality. By comparing the recorded acoustic signature of the whistle with an acoustic model of a theoretically perfect whistle, it is possible to determine a weighted set of deviations from the perfect model, those values are then combined to produce the final score.

Here's a flowchart of the basic analysis loop:
http://www.tradlessons.com/whistlemark/2.jpg

Display during analysis:
http://www.tradlessons.com/whistlemark/3.jpg

and a screenshot of the system in action:
http://www.tradlessons.com/whistlemark/4.jpg

Over the coming weeks, we'll be running a number of both cheap and expensive whistles through the system and will report the results. Based on what I've seen so far, I think there will be some real surprises.

Cheers,

Michael

[Loren]

Cool Michael.


Wouldn't it be funny if the "perfect" whistle will turned out to be a Recorder.....




Loren

[Dale]

I admit to being skeptical. More than a little.

The quality of a whistle includes a number of factors:

Quality of the materials. Durability, weight, etc.
Air flow, back pressure.
Workmanship.
Tonal quality.
etc.
etc.

You're proposing, if I understand, to measure one aspect of a whole constellation of elements that contribute to the quality of a whistle, it's tone characteristics quality. I've never been able to detect consensus about tonal quality of whistles. Some people like pure sound and some like some breath and noise.

What exactly is the "theoretically perfect whistle" ?

Maybe I'm missing something.

[Dale]

Cool Michael.


Wouldn't it be funny if the "perfect" whistle will turned out to be a Recorder.....




Loren

I wonder if any makers would be willing to volunteer to have their whistles measured to set the standard for the "perfect whistle."

[lalit]

Will a machine be playing the whistles? Otherwise, how do you account for player-introduced variation in tone quality? Average it for each whistle over a number of players? Use a human player and have her/him blow the whistle through an intermediary breath accelerometer? Are you measuring single tones, or is the software capable of a more multifaceted analysis?

It's nice to be able to measure whistle technology. It's interesting. It could be beneficial to makers and players. I don't think it will replace personal preference as a driver in how we like our whistles to play and what we like to hear in a whistle. I don't think there are absolutes in music. To me one of the beauties of music, whether from a scientific or artistic perspective, is that it arises as the negotiation of a complex series of compromises. It's inherently imperfect.

[Wanderer]

I dunno..on the one hand, we have player preferences, and there's no doubt that what you're used to has a great impact on what you think is proper. And then there's the subjective variable of taste (chiff/no chiff, etc).

On the other hand, we have some pretty strong mathematical and musical theory that should be able to be used to help show which whistles have more dissonant harmonics, etc.

I'm interested in seeing how this plays out.

[OBrien]

I think I feel my leg being pulled.

[Loren]

Cool Michael.


Wouldn't it be funny if the "perfect" whistle will turned out to be a Recorder.....




Loren

I wonder if any makers would be willing to volunteer to have their whistles measured to set the standard for the "perfect whistle."

Something tells me Dale that there will be no end of budding whistle builderers offering up their instruments as examples.

Loren

[ketida]

I think I feel my leg being pulled.

I agree. You simply can't disregard the highly personal exchange between the musician and any instrument...cheap, expensive, back pressure, or lack thereof, etc. There are just too many variables, including personalities.

That's something that cannot be scientifically measured, but makes all the difference in the end.

[CranberryDog]

Not THIS again! The problem with dealing with musicians is that they are musicians. Happy New Year! Cyril.

[ChaplainBlake]

This sort of mathematical model application to the judgment of standards in this case is akin to the application of ratios and symmetry to define beauty. I imagine that, in this case, like the case studies involving beauty, the data will be useful only insofar as generic "quality" is concerned. As a result, the data could be useful, but strictly limited in scope. In fact, this analysis would only be broadly useful insofar as we had the same abilities and same hearing/ear for music, and in the same venues to boot...and we do not.

In all reality, the only thing that this sort of exposition will tell us us how close a given whistle is the the theoretical "perfect" to which Mr Eskin refers. What is that "perfect"? As a singer, I would be hesitant to allow a computer model to establish the benchmark against which I would compare any singers. To carry the analogy further, would I allow a model to inform my opinion or perception of Rembrandt or Caravaggio, Bach or Palestrina? Never. And if a model formulated as the "perfect", a model from which standard deviation yields a raw score in an attempt to rank or establish quality, I would reject it outright. Bobby McFerrin's instrument is as fine as Pavarotti's, and any attempts to quantify the difference would be silly to any trained vocalist (other than remarks on their range or volume). Ever heard Palestrina in a grand cathedral? Ever seen a work of art that spoke to you? That is the sort of subjectivity that makes music inspiring and art compelling. Who is better: Ella Fitzgerald, Cecilia Bartoli, or Aretha Franklin? Good luck trying to solve that one with an applied mathematical formula.

There really is a point where data becomes useless in the practical sense, and this sort of endeavor seems to me to lead in that direction. As much as I seem opposed to such work though, I am actually intrigued by the principle (and will check the results). At any rate, it should spark some interesting conversations and add to others.

--Blake

[riverman]

Aw, C'mon guys, give Eskin a chance! It could be interesting!

...after all, it's not like you're going to ditch your favorite whistle just because the machines say its sound is less than perfect. And it could be helpful for low-experience whistle hacks like me who need the best whistle they can find just to sound half decent!

(assuming, of course, that I DO sound half decent!)

And we're whistlers! What ELSE are we going to write about? Laundry?

[rhulsey]

I think I feel my leg being pulled.

Aye, me thinks me chain's being yanked, as well.

But, I reserve judgment, and will gladly cooperate.

Reg

[FJohnSharp]

As soon as there is a computer program that can tell good poetry from bad, I'll jump in line for the whistle thing.

[PhilO]

I would love to see what he comes up with. I think in the end the results will bring varying responses from Are you kidding to I told you so

I once supervised a psychological screening program for a sensitive specific purpose and used a somewhat unique battery of tests; I was fortunate in that the tests actually agreed with my most important preconceptions about certain individuals. I think the degree to which the testing agrees with our preconceptions and experiences with whistles will determine how we respectively gauge the "Eskin Test"'s validity.

Happy New Year to all.

Philo