Does Old Cremonese sound exist - Subjective vs Measured

[Anders Buen]

Using a soft hammer sounds like cheating. And certainly trying to compare L values from impact data from non-identica equipment is questionable. In my one test where I compared my method to Curtin's calibrated rig on the same instrument, the L value from my measurement came out 4 dB higher.

Adjustments can also make a huge difference. Just recently I moved the soundpost on my #4 fiddle to try to reduce a strong peak in the transition hill. The L value before was 15.1, and after 20.0. The change was audible, and in the desired direction, but it didn't go from Chinese VSO to Strad. It was still mostly the same fiddle, just a little bit better.

Interesting data there Don. I think even the climate, that is the varying RH in air, might influence these numbers and the instrument response too, if the variation is large enough.

[Quentin Clark Jr]

Kreutzer Stradivari up for auction. Expected to bring around 7.5 million. Short (very short) video of Joshua Bell playing it at the auction house:

 

http://thestrad.com/latest/news/kreutzer-stradivarius-violin-to-be-auctioned-by-christie-s-in-new-york

OMG that thing is magical. Just a few seconds when it's miked to be heard but soooo worth it...

 

Thank you

[robertdo]

Would you get the same result if you were to test the same violin with several different strings? And do a Stradivarius or a Guarnerius sounds the same if it is played by Kogan or by Menuhin?

[Don Noon]

Then, Don,

Exactly what is it that you're trying to duplicate or to create?

Is it a certain set of parameters that can be expressed mathematically, in some manor?

Is it a sound or a tone, that may exist in your imagination, that you believe exists, but that (probably) you have not arrived at yet?

Have you ever played a violin that came really close, and you're trying, with your building, to duplicate that certain "ease of playing" or that "wonderful ability to create a certain full tone" that some particular violin exhibited?

Sorry, just seriously curious.

I do know how I'd answer that - and by the way, I've never played a Strad or a DG either.

 

The Dunnwald parameters are about the only mathematical attempt I know of to try and nail down what's good and bad.  I think it's much more complicated than that (isn't it all?), and I'm not really a fan of trying to boil everything down to a number.  However, I do think it's useful to look at details of the measured response to see what's going on, and how close you might be to getting there.  Impact response is a convenient snapshot, but lately I have been tending to look at bowed measurements as a more accurate (but difficult to read) picture.

 

So far, I have played 4 Strads and 1 DG, and of them I only thought one Strad was really impressive.  If you have followed anything I've been doing, you probably know it is the ex-Jackson, golden era Strad.  While it may not be my personal ideal tone for my own fiddling, I recognize its capabilities, and figure that if I can make one that performs like that, I'm happy (and then I can go on to make one that sounds better).  Needless to say, it hasn't happened quite yet.  Some acoustic features I think can be duplicated, others are more difficult, and getting them all in the same instrument is quite a lofty goal.  Here's the major pieces the way I currently see them, in terms of trying to make an instrument acoustically similar to a great Strad:

 

Signature mode frequencies:  relatively easy, and not very important as long as they're not way off.  Strad mode frequences aren't all that different from all other instruments, in general.

Signature mode amplitudes:  also no big difference between Strad  and decent modern, so it's not that big of a deal, and only matters for fullness.

Middle frequencies:  aaargh... very difficult to keep the overall level and peaks in this range down to a moderate level.  A major bane of new instruments, I think, giving strident notes on the E string (unpleasant) and a tone that sounds too buzzy and busy... not clean. 

High frequencies.:  High amplitude from ~1500 - 4000Hz, minimal dropouts, fairly steep cutoff above 4 or 5 kHz.  Getting even one of those features isn't something you can engineer directly; it's all part of the shape and materials in some complex way, and newly built instruments don't seem to be able to get there.

 

So, as I have been harping on for a while, shaping the middle and high frequencies I see as the main difficulty in making a great instrument, and age seems to be a factor.  There may be other aspects to consider (transients, bow response, under-ear sound), but it's a difficult enough task just to get the tone under control.

[Violadamore]

Interesting data there Don. I think even the climate, that is the varying RH in air, might influence these numbers and the instrument response too, if the variation is large enough.

It does.  Around here it's not unusual for RH to cycle from 40% to 100% and back over a 24 hour period (a month ago I was frequently seeing lows of 25% and highs of 90%!) and my violins change sound noticeably  along with it.

 

[Don Noon]

... lately I have been tending to look at bowed measurements as a more accurate (but difficult to read) picture.

 

 

Maybe it's not so difficult to read.

 

Here are low-resolution spectra of bowed semitone scales, which shows how the Jackson Strad differs from my #12 (VMAAI tone winner), #14 ("copy" of the Jackson), #15 and #16, both Strad models somewhat modified from the Jackson.

 

The Strad has a lower "transition hill" than any of mine, a broad, solid "bridge hill", and a sharp dropoff above 4 kHz.  My bridge hills do not look like that... either narrower, more uneven, or extending farther out into the higher frequencies, or several of these differences.  Of course, all of my instruments in this comparison are at most slightly over 1 year old, one is only a week or so.  So what happens over time?

 

I only have one older ("older" being more than 2 years old) instrument that is unmodified since I made it.  Comparing the impact spectra, it appears that the most change is an increase in the upper bridge hill area, and a slight filling in of some of the gaps (this fiddle was of processed wood, fyi).  So there is some hope that over time the bridge hill will re-form itself into a more Strad-like response.

[DMartin]

Don,

 

Thank you for these two illuminating posts and your underlying work.

 

Your statements fit well with my experience and current understanding.

Of old Cremonese, I have played only the Jackson and it's housemate dG.

 

Schleske, for one, has remarked that on better instruments the bridge hill resonances seem to organize in several somewhat rounded envelopes ( "tents" I think he said) of fairly equal height as suggested on your Strad low-res spectrum.  

 

So far,  I have not yet found a modern instrument with quite the clarity of the Jackson which I hear as bridge hill frequencies, though some get a long way in that direction. Will be interesting and frustrating if time is the necessary finishing ingredient.

 

My shop spectrum practice is mostly to bow an instrument with the software in free-running peak hold mode. I play as loud as possible through every fundamental frequency until the display stops growing in amplitude. Probably not valid, but seems to show the difference between instruments and to be repeatable enough. I also use the peak-hold mode for bridge tapping.

[curious1]

Don,

 

Thank you for these two illuminating posts and your underlying work.

 

My shop spectrum practice is mostly to bow an instrument with the software in free-running peak hold mode. I play as loud as possible through every fundamental frequency until the display stops growing in amplitude.

Does that mean you are playing a one octave scale/glissando starting with G (196hz)? Or are you playing every note on the violin?

[Peter K-G]

There is a lot of things that does not show up in bowing/playing spectra. It can look like it's great but it can still sound quite awful (I'm not going to post the badly recorded scale playing, they would be bullied out as horrific VSO violins)

 

 

I think it needs a lot of skills and experience, if not almost impossible, to recongnize Old Cremonese sound (if such does exist ) based on spectras. But I will continue to download clips and search on

 

What I have found and my way of expressing it, (pretty much the same things as above posts from Don)

Solid base (200  -600)

Some edge (800 -1300)

Brilliance (~2 KHz - 4KHz)

"Curved" cut off after 4 KHz => very important

[Craig Tucker]

Subjective vs measured.

... one seems to be as tenuous a "proof" as the other.

[Don Noon]

Subjective vs measured.

... one seems to be as tenuous a "proof" as the other.

 

The idea is to get agreement between the subjective and measured.  That would be more convincing, though still not real "proof" of anything.

[Peter K-G]

Subjective vs measured.

... one seems to be as tenuous a "proof" as the other.

 

 

The idea is to get agreement between the subjective and measured.  That would be more convincing, though still not real "proof" of anything.

 

With this I totally agree!

 

I have very little hope that we ever will find proof in this matter. To get somekind of agreement between subjective and measured would be a step forward.

 

Philosophically this puzzle is what's keeping everything about violins alive so we might not want the truth?

[Anders Buen]

A lot of the high frequency "noise" depend on how close one mics the recording, in what direction, probably with what rosin is used,, how damped the strings are, along with how close the fiddle is bowed to the bridge.. Many of the great violins sound quite harsh under the ears, somthing that is damped by the air absorption with distance. I think a violin will not enter the highest soloist level if this harshness is not there. Without it it might be a decent "blending in" fiddle, but nothing for a soloist.

Players prefer different instruments and there are many different uses of fiddles. So one "recipy" is not likely to work for all. It might work for those who happen to prefer the same as yourself, if one judges the quality by self playing it. A top maker can't rely only on his or her own ears only.

[DMartin]

Does that mean you are playing a one octave scale/glissando starting with G (196hz)? Or are you playing every note on the violin?

I'm trying to play every possible frequency on every string in order to discover and max out every resonance. Takes ten to fifteen minutes of torturing the instrument. This is just one kind of test. The results are repeatable enough for my purpose. Using the free-running, peak-hold mode allows one to see which bow actions triggered particular peaks as the spectrum takes shape over time. Hardly "scientific", and surely fraught with errors, but it does appear to confirm audible differences between instruments.

 

A single gliss on a string is a useful diagnostic but I find my glisses tend to miss resonances that take more deliberate bow action to initiate, particularly some high ones. I found this was also true of Oliver Rodger's gliss tests in his last year when he was very weak. I will be interested to find if better instruments show fewer "misses" in a weak gliss.

 

When I tap the bridge, I hold the fiddle in playing position and tap horizontally on the G-side, then over the top in between each pair of strings, and then on the E-side.

 

For what it's worth, I'm primarily interested in my emotional response. Numbers are interesting to me only to the degree they clarify issues—and I think they can.

 

I see myself as a messy person in a messy world. A computer plotted spectrum is useful only to the degree one knows exactly how it was produced.

[Marty Kasprzyk]

I'm trying to play every possible frequency on every string in order to discover and max out every resonance. Takes ten to fifteen minutes of torturing the instrument. This is just one kind of test. The results are repeatable enough for my purpose. Using the free-running, peak-hold mode allows one to see which bow actions triggered particular peaks as the spectrum takes shape over time. Hardly "scientific", and surely fraught with errors, but it does appear to confirm audible differences between instruments.

 

A single gliss on a string is a useful diagnostic but I find my glisses tend to miss resonances that take more deliberate bow action to initiate, particularly some high ones. I found this was also true of Oliver Rodger's gliss tests in his last year when he was very weak. I will be interested to find if better instruments show fewer "misses" in a weak gliss.

 

When I tap the bridge, I hold the fiddle in playing position and tap horizontally on the G-side, then over the top in between each pair of strings, and then on the E-side.

 

For what it's worth, I'm primarily interested in my emotional response. Numbers are interesting to me only to the degree they clarify issues—and I think they can.

 

I see myself as a messy person in a messy world. A computer plotted spectrum is useful only to the degree one knows exactly how it was produced.

Have any of you balsa violins been used in blind projection tests in comparison with Old Cremonese violins? 

[FlameMaple]

I have been using bowed response as well as impact spectra to try to get some quantitative documentation of violin tone, and there are definitely differences between impact and bowed response.  Then there is the problem of translating the hard data into how the player or listener react to the instrument.  While I do believe there are general things that can be said about all these relationships, one has to be careful about trying to say too much, as the pictures are not clear.  The impact spectrum has its use in that it is about as simple and clear a representation as we have about what an instrument does, even though it's far from perfect.  A low-quality picture of a ghost is still much more useful than a dozen conflicting descriptions by eye-witnesses.

 

Agree, Picture show the best perspective but eye-witnesses sometimes will make much perspective. Thank you for sharing

[DMartin]

Have any of you balsa violins been used in blind projection tests in comparison with Old Cremonese violins? 

Short answer is "no". They are too easily indentified.

 

However, it has been useful and humbling to hear many played in rotation with various old and new instruments and to also play those instruments myself.

[David Burgess]

Short answer is "no". They are too easily indentified.

 

However, it has been useful and humbling to hear many played in rotation with various old and new instruments and to also play those instruments myself.

Humbling for "more conventional" instruments, on occasion.

[Craig Tucker]

The idea is to get agreement between the subjective and measured.  That would be more convincing, though still not real "proof" of anything.

 

Yes, I must think that this is the truth of the matter. Convincing in a certain aspect, perhaps. If we could ever get an agreement. But still, no real proof.

 

As long as I have been interested in violins, there has been an on-going effort of somehow "classify" the subjective tone of Cremonese instruments, with a measurement of any sort - that other instruments do not (or, perhaps only on rare occasion) have.

All efforts to isolate anything of use, or of value, have essentially (in my opinion) failed.

As have most of the various "personal opinions" dealing with the subject. They remain highly suspect to me, as there is usually more emotion contained in them, than anything else.

 

Have we concluded anything yet?

(or does this thread simply go over, what everyone has gone over a thousand times before, with no solid answers?)

[Bruce Tai]

Last time we recorded 14 old Italian violin with a non-blinded violinist playing scales (5 Strads) 

I will instruct students to perform formant analysis working with numbered files in a blind way, and the instruments will be ranked by a program from 1 to 14. 

To pick top 5 to be all Strads by chance will be only 0.05%.

Even if 4 out of top 5 are Strads, the chances are only 2.25%.

If this experiment succeeds in picking out 4 Strads out of 5, would you guys agree that we have found one of the objective distinguishing features of Strads?

[David Burgess]

Won't the outcome largely rely on what the Strads are compared with?

 

It would be interesting to use instruments which soloists haven't (?) been able to distinguish from Strads in other tests.

 

Question: Were the Strads all adjusted by the same person or people?  If so, some distinguishable characteristic may be attributable to the preferences of these people.

[Craig Tucker]

 

If this experiment succeeds in picking out 4 Strads out of 5, would you guys agree that we have found one of the objective distinguishing features of Strads?

 

In a very real way Bruce, that would depend on certain things. First would be; what is your goal in performing this test?

If you have a certainty in either direction, that is, either for the idea that Strads can be distinguished by their tone , or; against the idea that Strads can be distinguished by their tone, then the results of a test (any test) designed by either you or anyone else that may be in the same "camp", would be not only suspect - but would eventually be torn apart by the people who would see whatever weakness the test had ( Dont believe me? see the posts about the any of the tests performed in the past to answer this - is it a Strad? -  question)

 

You'd have to take extreme caution that the test was unbiased in either direction, and that the people chosen to correlate the results weren't biased in either direction also.

It would help tremendously, if everyone involved in correlating the results of such a test, was not aware of what violins were played, or perhaps even, what the outcome was designed to show, so that any results were simply numeric results - with no specific connection to Strad or non-Strad violins.

 

Perhaps I'm being over zealous in my thinking... but I do know that all you will accomplish, if you are not exceedingly careful with the testing methodology and performance - is that you'll simply draw out the usually very intelligent and correct perceptions of interested parties, outside of the testing, that are going to show or "prove" that the testing was designed and performed to show one specific outcome...

 

Good Luck being unbiased ,and in performing a test to show what the fact of this matter is. 

I wish you the best, and am very interested in how you go about it, and exactly what results you get.

[Torbjörn Zethelius]

David,

I would think that the geometry of the internals of the Strads would give them away. I'm guessing that that's a large factor that make up the sound formants. It doesn't change with the set up. That, and the player, which is the one constant.

 

Psychology also plays a huge part. If you think that you can actually hear a difference, chances are better that you can do it as opposed to if you think you can't. Then you may just give up without hardly trying. Perhaps it's the largest factor of it all. That, and knowing what to listen for is very important. I believe that it takes practice to learn to recognize a certain instrument, but that it's possible.

 

Craig,

What I said above means that the test persons will have to be biased.

[David Burgess]

David,

I would think that the geometry of the internals of the Strads would give them away. I'm guessing that that's a large factor that make up the sound formants. It doesn't change with the set up. That, and the player, which is the one constant.

 

Doesn't the geometry of Strads vary quite a bit, and aren't there copies where the geometry is pretty close?

 

From Bruce Tai:

"If this experiment succeeds in picking out 4 Strads out of 5, would you guys agree that we have found one of the objective distinguishing features of Strads?"

 

I would consider it to be one piece of evidence, and not singularly conclusive, any more than I consider any other test to be singularly conclusive.

[Craig Tucker]

David,

I would think that the geometry of the internals of the Strads would give them away. I'm guessing that that's a large factor that make up the sound formants. It doesn't change with the set up. That, and the player, which is the one constant.

 

Psychology also plays a huge part. If you think that you can actually hear a difference, chances are better that you can do it as opposed to if you think you can't. Then you may just give up without hardly trying. Perhaps it's the largest factor of it all. That, and knowing what to listen for is very important. I believe that it takes practice to learn to recognize a certain instrument, but that it's possible.

 

Craig,

What I said above means that the test persons will have to be biased.

 

Well, the geometry of the internals of the Strads, in order to be the factor that would give them away - what would the test then consist of?

Looking at and measuring the internal geometry of all of the different violins in question?

On the other hand - there are many thousands of violins that have essentially the exact same ( within the bounds of objects existing in the physical universe) internal dimensions that Strads have.

The "dimensions" aren't necessarily the the difficult thing to duplicate. It's EVERYTHING, only (perhaps) coincidentally including the "exact" dimensions.

 

That they will have to be biased in the direction of finding out the truth? Rather than proving either bias , or, pre-conceived belief, is true?

Yes, in that case - I agree with you.