Jump to content
Maestronet Forums

Is the research on violin acoustics a viscious circle?


Andreas Preuss

Recommended Posts

2 hours ago, David Burgess said:

I think the emphasis now has shifted more in the direction of,
"If we have a better and more thorough understanding of how violins work, the greater the likelihood will be that we can tweak individual  characteristics as desired."

 No question that in some research the focus has changed. And imo it is the first time that all participants, violin makers, researchers, performers and  the audience, get some space to cooperate in solving the problem. However, I think that the idea of ‘tweaking individual characteristics as desired’ contains a kind of impossible mission. 
 

Science in general formed in us the the belief that we can control everything. We could also ask if we really do need to control everything. It is also thinkable that leaving certain aspects uncontrolled might turn out better results through an assumed self adjusting mechanism. This would actually resonate with the opinion of many skilled violinists, who are convinced in play-in-effects. And if we take this as real it is also clear that acoustic measure devices are not able to tell us what makes the play-in-effect. 
 

Provocatively one could ask: Doesn’t science create a hype for a total and absolute control, a goal which can’t be achieved?

Link to comment
Share on other sites

  • Replies 390
  • Created
  • Last Reply

Top Posters In This Topic

6 hours ago, sospiri said:

The famous quote may be applicable to some modern makers:
"If I have seen further, it is only by standing on the shoulders of giants."

I don't know of any contemporary pro maker who believes otherwise.
There's a lot to learn from examination of 300-to-400 years worth of instruments, both from what has been difficult to improve on, and also from changes made to accommodate  what has been learned in the meantime.

Example: My opinion is that the switchover from the nailed surface-mounted neck to the mortised neck, had mostly to do with the eventual realization that the neck projection on pretty much all violins changes over time, and that it was a lot easier and less destructive to change the angle on a mortised neck, than on a nailed neck.

Link to comment
Share on other sites

9 minutes ago, Andreas Preuss said:

No question that in some research the focus has changed. And imo it is the first time that all participants, violin makers, researchers, performers and  the audience, get some space to cooperate in solving the problem.

The various Oberlin workshops have been collaborating that way for at least 20 years, with the assistance of high-level players.

I do sometimes get a little frustrated when someone tries to "reinvent the wheel", without having examined "prior art" (as it's called in the patent business).

Link to comment
Share on other sites

1 hour ago, Andreas Preuss said:

Science in general formed in us the the belief that we can control everything.

That's like saying a mirror is responsible for the way you look.

1 hour ago, Andreas Preuss said:

Provocatively one could ask: Doesn’t science create a hype for a total and absolute control, a goal which can’t be achieved?

Daedalus warned Icarus to fly neither too high nor too low.  It was Icarus who ignored him.  'According to scholia on Euripides, Icarus fashioned himself greater than Helios, the Sun himself, [Wiki]'  Again you blame the mirror for it's reflection?  One of the reasons I like Don Noon - he restricts himself pretty much to describing the image.

4 hours ago, David Burgess said:

That created quite a dilemma: "Since it has turned out that there isn't a specific type of sound to recreate, what do we do, now?" default_ohmy.png

The first step is always going to be to define good and bad violins.  If we can't do that even for extreme cases then we're dead in the water.  If we can agree for extreme cases (and throw out the 'maybes' for this experiment), then instrument them the same way as closely as possible and collect the data.  Repeat the same process with the same violins enough times to get a meaningful average for each group.  If the groups can't be differentiated on the basis of the data, then we need to rethink the experiment.  If they can be differentiated then design experiment # 2 (TBD).

Let's say we have two groups of pencils: same color, same hardness, all with unused erasers.  Which pencil belongs to which group may be difficult to differentiate based on, say, exposed eraser length, diameter, ferrule size, fade marks on the paint etc. no matter how accurate and precise our measuring devices are.  Some wag may point out that one group is two inches longer than the other and problem solved, but maybe not and if no-one can reliably differentiate them, then maybe there aren't two groups.

Link to comment
Share on other sites

1 hour ago, David Burgess said:

The various Oberlin workshops have been collaborating that way for at least 20 years, with the assistance of high-level players.

I do sometimes get a little frustrated when someone tries to "reinvent the wheel", without having examined "prior art" (as it's called in the patent business).

I am not trying to reinvent the wheel in contrary. It’s just about getting away from the solid wood wheel…IMG_0560.thumb.jpeg.7fa4313dfb4dd24a53e733ca4ef64c52.jpeg

we are a kind of still at the wheel on the left and are trying to make the circle perfect. And actually I think this demonstrates in a simple analogy where we could go. With the invention of spokes, weight could be saved. 

Link to comment
Share on other sites

7 hours ago, sospiri said:

The evolution of the violin happened long before 'scientific research'. So I disagree with your points here.

The evolution has nothing to do with quality. Back in the day the cremonese makers were the best as they were the guys who invented the thing and produced it at high levels. Other makers of the time only emulated their instruments with mixd results both visually and tonally from what they saw without knowing the building methods until the time when Amati took aprentices and several new maker families started working at the high level of Amatis. Later the mass production and other social/economic factors caused that for long time quality of violin making was on decline so these old instruments were the best available by a margin till makers like Hills etc. started revealing the old construction methods and the best makers were able to produce instruments close to the originlals, but by the time the old instruments were already settled on the throne (with help of clever dealers who had finantial interest in driving prices up and invented all kinds of marketing stories about magical varnish that is impossible to recreate etc.) and till now that kind of marketing prevents other makers to get close to the status unless he makes such a fake that the conoisseurs cannot tell from original.

These days we have all the information about construction methods and varnishes used available and selection of best woods and variety of perfect strings so there is no reason to believe the best makers of today (which are plentiful compared to the few guys around Cremona in 17th century) cannot be as good as the old guys.

Re. the acoustic research. I highly doubt that the research that analyzes direct connection between violin dymnamics and tone will get results other than perhaps understanding wolf tones and how to get rid of them or helping with tonal adjustments via soundpost or bass bar position/shape. The violin is way too complicated system for such model ananlyses. I would expect that direct application of AI being the step in right direction (and I wrote this year or two ago, well before the AI boom) in analysis of vioin or at least finding what parameters define and shape the tone in desirable way.

Link to comment
Share on other sites

1 hour ago, Andreas Preuss said:

I am not trying to reinvent the wheel in contrary. It’s just about getting away from the solid wood wheel…IMG_0560.thumb.jpeg.7fa4313dfb4dd24a53e733ca4ef64c52.jpeg

we are a kind of still at the wheel on the left and are trying to make the circle perfect. And actually I think this demonstrates in a simple analogy where we could go. With the invention of spokes, weight could be saved. 

LOL, "Reinvent the wheel" is a metaphor, a "figure of speech", a colloquialism which I have never experienced being used with the intent that it be taken literally. ;)
What I was trying to put across is that the collaboration between violin makers, researchers, performers and audiences which you suggested, has already been going on for quite some time. 
The "Fritz et al" experiments are only one of the more recent, better documented, and more easily available examples.

Link to comment
Share on other sites

21 minutes ago, Dr. Mark said:

The first step is always going to be to define good and bad violins.  If we can't do that even for extreme cases then we're dead in the water.  If we can agree for extreme cases (and throw out the 'maybes' for this experiment), then instrument them the same way as closely as possible and collect the data.  Repeat the same process with the same violins enough times to get a meaningful average for each group.  If the groups can't be differentiated on the basis of the data, then we need to rethink the experiment.  If they can be differentiated then design experiment # 2 (TBD).

Let's say we have two groups of pencils: same color, same hardness, all with unused erasers.  Which pencil belongs to which group may be difficult to differentiate based on, say, exposed eraser length, diameter, ferrule size, fade marks on the paint etc. no matter how accurate and precise our measuring devices are.  Some wag may point out that one group is two inches longer than the other and problem solved, but maybe not and if no-one can reliably differentiate them, then maybe there aren't two groups.

There are math models that can help but they need more data than one single luthier can give in his own lifetime (unless he has as busy workshop as Stradivari). The AI algorithms may be more effective with data that may appear random but also require large pool of data to learn from. Perhaps mutual collection of data provided by best luthiers could be used to feed AI.

Link to comment
Share on other sites

Serious question: are there any real modern and informed violin makers who make vioilns that are definitely bad? Aren't they all just variations on good, except for the occasional odd one? So why is it necessary to define the two at all. Just assume good, and try to separate out what makes them different from each other. If you have enough bad violins that you need technology to fix what you're doing, the problem should be identifiable and fixable without technology, shouldn't it?

Link to comment
Share on other sites

1 hour ago, Michael Darnton said:

If you have enough bad violins that you need technology to fix what you're doing, the problem should be identifiable and fixable without technology, shouldn't it?

Yup, but I don't think that making good violins out of bad violins is the current pursuit or main focus. I think it has more to do with the hope of expanding the potential options and possibilities beyond what most makers have readily available to them now.

For example, if response could be improved beyond what's available now, without compromising anything else, I think a lot of people would be very interested in that. And some might even consider it to be a "superior" violin. Others might not, if the response differs too much from what they are accustomed to, or "plays different".

"Mixed reviews" were kinda what happened when the first really successful synthetic strings, Dominants, came on the market. I remember that a lot of people really hated them. A few years later, many of these same people had switched over, and were using them exclusively.
Now, the use of gut-core strings has gotten rather uncommon.

Link to comment
Share on other sites

43 minutes ago, HoGo said:

The AI algorithms may be more effective with data

AI and ML are certainly tools we can try to use to meaningfully classify recognized groups of violin tone.  Trying to learn from AI results though can be a royal pain in the tokus.  AI has a tendency to do odd things - like create vectors containing bins of correlated frequency amplitudes, and when we look at them closely we find that every single violin is individually and independently represented by a long string of weighted bits representing a fraction of this frequency amplitude and a dash of that frequency amplitude.  If we're lucky, new data also causes output 1 to be high and 2 to be low when it should.  If we're not lucky, AI may tell us that it's a viola.  On the surface it doesn't look like grandmothering, but the number of individual data vectors that can be stored in AI is immense.  We just label the experiment a success, publish, grab next year's funding, and move on.

4 minutes ago, Michael Darnton said:

Serious question: are there any real modern and informed violin makers who make vioilns that are definitely bad? Aren't they all just variations on good, except for the occasional odd one? [>>>]So why is it necessary to define the two at all. Just assume good, and try to separate out what makes them different from each other.[<<<] If you have enough bad violins that you need technology to fix what you're doing, the problem should be identifiable and fixable without technology, shouldn't it?

Say we make it easy - differentiate only by very dark vs. very bright tone (there he goes again, thinks my psychologist).  To start with we exclude those that have both characteristics and those that we can't agree on.  Get a sufficient number of sufficiently influential people to agree about what violins belong in each group.  Maybe we could do that.  Take some acoustic measurements, say a psd so we don't need to worry about phase, with experimental conditions as near as possible to identical each time, and only one moderately far-field set-up to start with since we don't need the complications of multiple positions and orientations and stuff like that yet.  Rinse and repeat a few times and average the results for each violin.  Now if things go well we have some data.  The best first step in analysis, if possible, is to visually scan the two datasets for differences correlated with initial class.  With luck something simple will pop out.  If not then we think of ways to present the data such that existing differences will pop out - the usual ways, or maybe some advanced standard analysis like a cepstrum, or something.  If this gets difficult we become obsessive-compulsive.  We'd have to see the data...

Let's say we succeed, because we're really, really good at what we do.  Well, now we found some acoustic characteristics that correlate with our tonal classification.  What next?

Now we look at the violins in each group - the material properties, dimensions, varnish etc etc.  With luck we can find unique physical characteristics that differentiate the groups, i.e. correspond to the different tonal characteristics.  If we do we can build a bright or dark violin.  One may ask why we need the acoustic measurements if we can do this?

[>>>] Two distinct classes seems to be the easiest way to get general agreement on the quality of the violins, and also is the simplest differentiation experiment I can think of that might have some reasonably unambiguous result.  If we can't measure anything that differentiates two pretty extreme classes then we have a problem.  If we can, then we may have insight into what to look for in less extreme cases.

I'm mostly using as many words as possible to propose picking the low hanging fruit first.  Succeed with the easy stuff and the hard stuff becomes easier.

 

 

 

 

Link to comment
Share on other sites

I remember Oliver Rodgers' work in this. If I remember this correctly he took measurements of a couple of years of VSA contest winners. Everyone agreed that they were good. Then he did some Strads and similar violins. I think people probably agreed they were good. The problem was that his FFT graphs of the two groups were completely different. And neither set dealt with the dynamic changes under playing conditions in the two types, which I suspect are more important and also quite different. 

I think picking a violin might turn out to be like picking a spouse. You think you should be able to get it down to, say, a simple checklist of 5,000 to 10,000 multiple choice questions but you still won't be able to predict personal chemistry.

 

Link to comment
Share on other sites

Oliver Rodgers was using bowed glisandi, picked up by a microphone. I'n not sure an amateur players bowing is repeatable enough to reliably and repeatedly extract fine detail. Even with a professional's bowing, there are repeatability problems. That's the main reason Norman Pickering came up with a bowing machine, and I think that's one of the main reasons so much of the testing has moved to calibrated impact hammers.

I don't recall seeing Rodgers comparison of VSA winners with Strads, so can't comment on that.

Link to comment
Share on other sites

When his software became available I bought it and used it for a couple of years. Theory aside, my results with it were quite consistent. That period for me lasted about 10 years through three different programs, and while useful for helping me learn how to listen critically it didn't have much effect on my making. Generally I found the stability of violins was sufficient that my bowing didn't obscure the differences between them.

 

Link to comment
Share on other sites

Oh, in between his use of a bowing machine, and things switching more to impact hammers, Pickering was using a phonograph record cutting head to drive the violin with a sweep at quarter-tone(?) increments. The main issue was that the cutter head added so so much mass to the instrument, that mode frequencies were altered from where they would be without the cutter head. But it was useful for comparing different violins, as long as the same rig was used.

Impact hammers were much lighter, and would spend very little time in contact with the instrument, so they had much less influence on frequencies, damping and mass.

Testing methods keep being further refined.

Link to comment
Share on other sites

I wouldn't trust the testing.  Even when you'd think a researcher should be qualified, the field attracts eccentrics and he's likely as not to discover the secret of cold fusion while violin testing.  Just an impression.  The best bet would be some disinterested, music hating if possible, signal processing PhD hired away from the semiconductor industry for a month

Link to comment
Share on other sites

If he could get effectively identical results from a control violin before each test violin was examined then he may have had sufficient control over the tests, even if run at different times and places.  That would entail repeated measurements with similar results from the reference violin, before testing each Strad or VSA violin.  If he tried this and the control violin responses were too noisy then he was wasting his time.  Otherwise if he recorded the responses of the control violin each time then he could have deconvolved the differences in control response to normalize the test data and probably get something meaningful.  Not that anything would be observable if there are no significant tonal differences, but that's a different issue.

Wow this thread has legs.

Link to comment
Share on other sites

54 minutes ago, Bill Merkel said:

I wouldn't trust the testing.  Even when you'd think a researcher should be qualified, the field attracts eccentrics and he's likely as not to discover the secret of cold fusion as a byproduct of violin testing.  Just an impression.  Some disinterested, music hating if possible, signal processing PhD hired away from the semiconductor industry for a month would be the best bet...

Even Davis, who has been highly involved in the Oberlin Acoustics workshops, is/was a Boeing Aircraft vibration analyst engineer. George Bissinger, who has also been highly involved, is/was a nuclear physicist. He doesn't make or play violins, to the best of my knowledge. Claudia Fritz, also very involved, is a PhD research scientist who doesn't make or play violins. Do you think they might be OK?

Link to comment
Share on other sites

10 minutes ago, Bill Merkel said:

No.  Well, yes, if you want the violins to fly or undergo fission.  It's a signal processing problem, quite a specialty

I don't think Evan Davis was involved in getting things to fly, but in making sure aircraft and their components wouldn't vibrate or oscillate themselves apart during flight, which neither airlines nor their passengers seem to enjoy. ;)

Link to comment
Share on other sites

58 minutes ago, Bill Merkel said:

Would you want your aerospace guy doing the nuclear research?  The nuclear guy making sure your wings didn't fall off?  But their signal processing is tops I'm sure

I'm pretty sure that both involve a lot of signal processing. What would you be looking for? A chip designer? A software engineer? A radar array designer?

Pickering was the guy who invented/designed the Pickering phonograph cartridge, which  you may remember if you're old enough. Later, he was a string designer/engineer for D'ddario. He was the guy who got me started on acoustical research, and also the guy who told me that human bowing wasn't gonna cut it, if one wanted to pick out the fine details. Often, the difference between two bow strokes, even when a  professional was trying to do them exactly the same, could be bigger than the differences between two violins.

Link to comment
Share on other sites

18 minutes ago, Bill Merkel said:

Somebody with a background in amplifiers with specialized physics knowledge of mechanical analogies who makes a peanut butter sandwich using differential equations

I bet they could bring one in if they thought it would be helpful. In what way do you think their current signal processing might be coming up short, and why do you think so?

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.



×
×
  • Create New...