Jump to content


Photo

Critical frequency and signature modes


  • Please log in to reply
57 replies to this topic

#1 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 07 July 2011 - 09:45 AM

I have been playing with some data from one of Bissingers articles from JASA in 2008: Structural acoustics of good and bad violins JASA 2008
One of the results or claims there are that better instruments have a lower critical frequency. The critical frequency is the frequency where the bending waves in the wood and the speed of sound in air is the same. The radiation is then peaking in its efficiency. We see this quite clearly in sound reduction measurements and impact noise measurements in the architectural acoustics where the sound reduction curve has a dip and the impact noise curve has a peak in the frequency response. But the effect is not so easy to spot in spectra from violins.

Anyway, I have copied Bissingers data from his figure where he compare the subjective rating and the measured critical frequency (his definition of the critical frequency is a bit different from the classic way, he uses the point for where the regression fitted radiation efficiency fitted line crosses the 0dB line (or 1 in linear scale. The classic critical frequency will lie higher than that by some 1kHz or so)

I enclose the figure with my copied data above. As you see he has grouped the scores into three classes, and he use the average radiation efficiency lines for the bad versus the good and excellent to reach the conclucion that the better violins tend to have lower critical frequencies.

I have tested the data including all of the points for critical frequency against the subjective scores and tested if there is a significant correlation there, and thus: if the slope between subjective rating and critical frequency is significantly different from zero. It turns out it isn't. There is a slope for the regression line, but the slope is not statistically significant.

The correlation coefficient is -0.47, but the p-value is 0,2. If the p-value had been below some 0.05 there would have been a signal there.

A lower critical frequency would idicate a thicker top plate, stiffer wood, or both.

Attached Thumbnails

  • Critical frequency versus subjective rating.jpg
  • Critical frequency vs subjective scores with regression fit.jpg

"If the facts don't fit the theory, change the facts". A Einstein

#2 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 07 July 2011 - 10:00 AM

I have also had a look into the signature mode data in the same article and copied them for a statistical analysis. Bissinger say that there are no correlation between the signature mode frequencies and subjective quality as rated by his violinist.

I could find a weak trend for the B1- data. The p-value is 7%, so it is a bit too high for a clear conclucion. For the B1+ frequency the trend is too weak and the other modes show nothing.

The average B1- frequency in Bissigers data are 471 +/- 17Hz (+/- one standard deviation) which is pretty high. In my own database including data for 46 old Italian violins (mainly Curtins data) the average is 444 +/- 15Hz which is significantly lower than Bissingers data. Either the climate in his laboratory must be very dry, or the violins in the set are on the stiff side.

In my Cumulative Distribution Function plot the green data points are data from Old Italian violins (plus a few top contemporary makers), the red ones are violins passing my shop and the black are Hardagner fiddles that has passed my shop.

The correlation between B1- frequency and subjective quality is not crystal clear, but I think there might be a trend there.

Attached Thumbnails

  • Signature modes and subjective rating.jpg
  • Signature modes and subjective rating with regression line for the B1-.jpg
  • B1- data from my own set + 46 old Italians.jpg

"If the facts don't fit the theory, change the facts". A Einstein

#3 JimMurphy

JimMurphy

    Enthusiast

  • Members
  • 1792 posts
  • LocationUSA

Posted 07 July 2011 - 10:17 AM

Hi Anders,

Interesting data & graphs.

With respect to "Subjective Quality ratings", my #1 technical [and 'objective'] question is: Given different sets of strings impart different tension [etc.] on the resonant instrument [violin], did ALL tested violins use the same Brand & grade of strings, or wasn't that considered important?? :o

Thanks,

Jim
"To see things in the seed, that is genius." ~ Lao Tzu

#4 Wm. Johnston

Wm. Johnston

    Enthusiast

  • Members
  • 2042 posts
  • LocationSouthwest Virginia

Posted 07 July 2011 - 11:32 AM

Anyway, I have copied Bissingers data from his figure where he compare the subjective rating and the measured critical frequency (his definition of the critical frequency is a bit different from the classic way, he uses the point for where the regression fitted radiation efficiency fitted line crosses the 0dB line (or 1 in linear scale. The classic critical frequency will lie higher than that by some 1kHz or so)

I haven't looked at this critical frequency before. Combining the numbers on the plots with the +1kHz in frequency to get to the classical critical frequency, that puts this number around the extra bump in the spectrum that Curtin wrote about in his Vieuxtemps Guarneri article. Do you think there could be a relation between these? I've been testing all of my violins with vertical excitation, because that is the easiest way to excite a violin when it doesn't have a bridge, and mine seem to have a hill in the response around that region too.
Website
Available Instruments
Not a member of the magic potion school of violinmaking.

#5 Don Noon

Don Noon

    Using tools without supervision

  • Members
  • 5049 posts
  • LocationCarlsbad, CA

Posted 07 July 2011 - 12:16 PM

I think that a great deal of possibly useful information gets vaporized when you try to boil things down to a single number such as a "subjective quality rating". Violins can be good or bad for many different reasons, and selected critera, such as signature modes and critical frequency, may (I'd say probably) have very discernable effects on tone that get homogonized in the rating process.

Then there's the problem of "critical frequency", where you can get a good value by using crappy wood 7mm thick, or really fabulous wood of normal thickness. I think you'd get wildly different quality ratings, even if the critical frequency was identical.

My guess is that there would be much more useful information for violin makers if these types of survyes were done with a little more granularity... breaking down the ratings into things like dark, bright, power, clarity, and balance... where there might be some more clear correlations.

Making fiddles ain't rocket science...  it's much more complicated.


#6 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 07 July 2011 - 01:32 PM

Do you think there could be a relation between these? I've been testing all of my violins with vertical excitation, because that is the easiest way to excite a violin when it doesn't have a bridge, and mine seem to have a hill in the response around that region too.

Could be. Another possible way to see the critical frequency could be to see where the response increase when the violin is damped. Havent looked into that in detail yet.
"If the facts don't fit the theory, change the facts". A Einstein

#7 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 07 July 2011 - 01:39 PM

I think that a great deal of possibly useful information gets vaporized when you try to boil things down to a single number such as a "subjective quality" wildly different quality ratings.

I think he has that kind of information. He has an CASJ article on the questionaire made for the purpose. It was quite elaborate. We do rate violins and are able to quite cosistently, but we do not agree on the rating. So scores will be individual to our own preferences.

This is a piece in the puzzle.
"If the facts don't fit the theory, change the facts". A Einstein

#8 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 07 July 2011 - 06:04 PM

Hi Anders,

Interesting data & graphs.

With respect to "Subjective Quality ratings", my #1 technical [and 'objective'] question is: Given different sets of strings impart different tension [etc.] on the resonant instrument [violin], did ALL tested violins use the same Brand & grade of strings, or wasn't that considered important?? :o

Thanks,

Jim

The critical and signature mode frequencies are independant on the strings. But the subjective rating might depend on the strings. This is not my data set and have not been collected by me directly, so I do not know which strings were on these fiddles. However, I know and have played a few of them, e.g. the "David Rubio sixlings" which I do think had Dominants on them. The three Strad 3D instruments are also a part of the set, so maybe the pics in that DVD will tell what strings were on these? I do not think the three finest violins in the data set were subjected to the same rigorous quality rating as the other 14 instruments.
"If the facts don't fit the theory, change the facts". A Einstein

#9 JimMurphy

JimMurphy

    Enthusiast

  • Members
  • 1792 posts
  • LocationUSA

Posted 07 July 2011 - 10:53 PM

The critical and signature mode frequencies are independant on the strings. But the subjective rating might depend on the strings.

Hi Anders,

I was thinking Bissinger's critical frequency was derived from "radiation efficiency" data which may be altered by different tensioned strings [than on other tested violins]. Don't know for sure though. And indeed the player's "subjective rating" would be influenced by different strings.

I expect it's tough enough just getting all the violins together for test purposes. I imagine there are many owners who'd be very fussy about having strings changed for a test, making it more difficult to standardize the test procedure [with like strings].

Jim
"To see things in the seed, that is genius." ~ Lao Tzu

#10 jezzupe

jezzupe

    Grand Poobah

  • Members
  • 2839 posts
  • Locationbela marina

Posted 08 July 2011 - 12:04 AM

With these kinds of test's the "good vs bad" I think is where the problem starts. I think those of us who hear the violin differently than the "average' listener still have a major stumbling block. That is what Don was eluding to. The "flavor" of the sound. Until there is a general consensus on what different "flavors" defined sound qualities are, I thnk we are all blind men trying to describe what part of the elephant we are holding. I think in this day and age it would be a good idea for a "flavor" guide to be established by some qualified group related to defining sound quality and just what "we" mean when we say ...dark or sweet or twangy etc.

I would think that it would be helpful to all parties involved, particularly for newcomers, if, for example, a recording catalog were compiled of various violins that have agreed on tone qualities. For example, " this is the "dark" category...the following 5 recordings are of violins that have a general consensus of several established luthiers/players of having this sound quality. The following qualities x,y,z are what qualities within the general tone quality that define this tone description"....something to that effect.

One of the things that I do, quite inadvertently, is create instruments that all have very dissimilar sound from one another. They all have a very individual sound, and even with my own instruments I sometimes have a hard time describing what flavor they are as I now have so many that one tends to judge one based on the one they just heard. I can come back latter and play a different one first, and then the same comparison one, and it will have a different quality because I am now comparing it to something else.

I thik most of us know when we hear a bad violin, I generally describe them as 'dead,flat,tinny,quite,muted' etc. And I do think we should have samples so one can hear the difference. Perhaps some day some of us hear could work on a tone/flavor guide.

Because I do know that I have heard violins that I hated, that certain pretty well known players loved, generaly non classical players who like a sound that would not work well in a orchestra, and of course the other way around.

#11 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 08 July 2011 - 02:08 AM

Hi Anders,

I was thinking Bissinger's critical frequency was derived from "radiation efficiency" data which may be altered by different tensioned strings [than on other tested violins]. Don't know for sure though. And indeed the player's "subjective rating" would be influenced by different strings.

I expect it's tough enough just getting all the violins together for test purposes. I imagine there are many owners who'd be very fussy about having strings changed for a test, making it more difficult to standardize the test procedure [with like strings].

Jim

I do not think the strings has an effect on the critical frequency or the radiation efficiency, as already stated. It is a physical property related to the plates, body and wood alone.

Strings are not that important to a violins overall quality, but does matter when not good strings are used. I do not think any bad strings have been used in these tests. Mostly Dominants, I guess.
"If the facts don't fit the theory, change the facts". A Einstein

#12 carl stross

carl stross

    Enthusiast

  • Members
  • 3297 posts

Posted 08 July 2011 - 04:16 AM

I do not think the strings has an effect on the critical frequency or the radiation efficiency, as already stated. It is a physical property related to the plates, body and wood alone.

Strings are not that important to a violins overall quality, but does matter when not good strings are used. I do not think any bad strings have been used in these tests. Mostly Dominants, I guess.


Anders, strings are extremely important to a violin's overall quality. A suitable change of strings can turn a "plank" into a passable instrument provided the player will put up with them. Cellists are far more prepared to do this.

I think what Bissinger does is somehow too wide to be of practical importance ( besides writting "papers" ).

#13 lyndon

lyndon

    still trying to post #4090

  • Members
  • 4089 posts
  • LocationWest of Cremona, East of Beijing

Posted 08 July 2011 - 04:34 AM

i think the frequencies a violin vibrates at wouldnt change much with different strings, the quality of the tone at those frequencies could change a lot though with different strings
Taylor's Fine Violins, Redlands, S. California
Specializing in the research and restoration
of baroque, transitional, and modern violins.

http://www.violinist..._johann_taylor/
(violin shop ad, with links to instruments for sale, pictures of
violins I restored, and recordings and pics of my clavichords)

#14 jezzupe

jezzupe

    Grand Poobah

  • Members
  • 2839 posts
  • Locationbela marina

Posted 08 July 2011 - 04:40 AM

i think the frequencies a violin vibrates at wouldnt change much with different strings, the quality of the tone at those frequencies could change a lot though with different strings

yup

#15 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 08 July 2011 - 08:06 AM

Anders, strings are extremely important to a violin's overall quality. A suitable change of strings can turn a "plank" into a passable instrument provided the player will put up with them. Cellists are far more prepared to do this.

I think what Bissinger does is somehow too wide to be of practical importance ( besides writting "papers" ).

Well I do not experiment much with strings. On violins I put on Dominants and a Kaplan E or a Wondertone or something similar. On Hardanger fiddles I may change the gauge if there are a weakness or too much power on a string. These changes does not change the character of the instrument, but makes it sound more even.

I know how the chinese steel strings that comes with the strung up instruments sound and feel like and what Dominats sound and feel like in comparison. I know that difference and prefer the Dominants. I have also played Zyex blue. The character of the instrument does not change that much with the strings. But you may have more experience on the matter.

I am not ready to spend 100ds of dollars in investigating the properties of different strings, as what I do to instruments are mostly related to their body and fittings. I have tested different gouge Dominants to investigate how that ma effect long time average played spectra and how they play. But at that time I did not find anything revolutionary, just that the sound might be stronger with the heavier set.

I may dig into this at a later time, but not now. Do anyone have recorded material or documantation to share on the issue? I would be very interested to see that.
"If the facts don't fit the theory, change the facts". A Einstein

#16 carl stross

carl stross

    Enthusiast

  • Members
  • 3297 posts

Posted 08 July 2011 - 08:20 AM

Well I do not experiment much with strings. On violins I put on Dominants and a Kaplan E or a Wondertone or something similar. On Hardanger fiddles I may change the gauge if there are a weakness or too much power on a string. These changes does not change the character of the instrument, but makes it sound more even.

I know how the chinese steel strings that comes with the strung up instruments sound and feel like and what Dominats sound and feel like in comparison. I know that difference and prefer the Dominants. I have also played Zyex blue. The character of the instrument does not change that much with the strings. But you may have more experience on the matter.

I am not ready to spend 100ds of dollars in investigating the properties of different strings, as what I do to instruments are mostly related to their body and fittings. I have tested different gouge Dominants to investigate how that ma effect long time average played spectra and how they play. But at that time I did not find anything revolutionary, just that the sound might be stronger with the heavier set.

I may dig into this at a later time, but not now. Do anyone have recorded material or documantation to share on the issue? I would be very interested to see that.



Yes, the cost is prohibitive but you could also keep only a couple of sets for testing purposes. Specially on G,D the string is fundamental for what the listener hears and I'm sure this would reflect in your average spectras. My observation was that a heavier string makes not only for a more powerfull sound but also for a more complex one. Of course at some cost to playability.

#17 JimMurphy

JimMurphy

    Enthusiast

  • Members
  • 1792 posts
  • LocationUSA

Posted 08 July 2011 - 08:21 AM

I do not think the strings has an effect on the critical frequency or the radiation efficiency, as already stated. It is a physical property related to the plates, body and wood alone.

Bissinger's 'calculated' [rather than 'measured'] critical frequency is a fudged number derived in part from plate bending behavior which IS dependent on string tension. The mathematician in me is a bit leery of drawing objective conclusions using fudged numbers especially if all violins weren't under the same string tensions. I won't belabor the point any further. It just bothers me though.

Thanks,

Jim
"To see things in the seed, that is genius." ~ Lao Tzu

#18 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 08 July 2011 - 08:32 AM

I think what Bissinger does is somehow too wide to be of practical importance ( besides writting "papers" ).

Could be. But at least he does provide some ideas into what may play a role for violin quality and how the violin works in general. It is difficult to follow his articles, even for a specialist. E.g. the 3D Strad project would not have been without his lab and cooperation.

E.g. he say that the A0 level was stronger in the Old Italians he has measured in comparison to the contemporary instruments he has data from. I think that may include a Curtin and a Zyg violin, as well.

Bissinger has written an extensive amount of literature on the violin. Lately he even has read former reserchers work and digested it in an effort to come up with useful material amd model for makers, in his JASA article Structural model of the violin radiativity model from 2009. He has been part of the VSA Oberlin violin acoustics teacher team for many years, and for some years he also brought his lab to the workshop. They did a large bridge tuning experiment in 2005 of which he has reported in the JASA.

His literature seem to be aimed at a scientific community. But ha has shared all his ideas with makers at the Oberlin workshop. And I think he now is working witout a grant or any financial support. He is genuinly interested in the subject.

Personally I have been interested in the critical frequency issue, as it comes from the architectural acoustics literature and is something we see daily in my daytime job environment. Cremer write about it in his book on the violin, and possibly was one of the inventors of the subject. I have written a draft for a paper on it some 7 years ago now which includes the radiation efficiency and average plate response. I haven't digged into the issue as I could have with alternative measurements and investigations. One of my collegues was a PhD student under Frank Fahy which Bissinger also has worked with and used as a source.

With a larger dataset he could have reached more sound and robust conclucions. This far he seem to confirm some of Dünnwalds findings, e.g. the weakened 650-1,12kHz range in the better rated instruments and the stronger A0.
"If the facts don't fit the theory, change the facts". A Einstein

#19 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 08 July 2011 - 08:39 AM

Bissinger's 'calculated' [rather than 'measured'] critical frequency is a fudged number derived in part from plate bending behavior which IS dependent on string tension. The mathematician in me is a bit leery of drawing objective conclusions using fudged numbers especially if all violins weren't under the same string tensions. I won't belabor the point any further. It just bothers me though.

Thanks,

Jim

No Jim. He only look at the average plate vibration velocity and the radiated sound in quite wide frequency bands which are NOT dependant on the string tension.

If I hadn't told you that his critical frequency is different from the definition of it, you would not have started to talk about "fudged numbers". I think you are "bluffing your way in matematics" here. :-)
"If the facts don't fit the theory, change the facts". A Einstein

#20 Anders Buen

Anders Buen

    Enthusiast

  • Members
  • 4290 posts
  • LocationOslo, Norway

Posted 08 July 2011 - 08:43 AM

Yes, the cost is prohibitive but you could also keep only a couple of sets for testing purposes. Specially on G,D the string is fundamental for what the listener hears and I'm sure this would reflect in your average spectras. My observation was that a heavier string makes not only for a more powerfull sound but also for a more complex one. Of course at some cost to playability.

Well, I do have a very good control on the D string sound, and to a large extent also the G without need for varying strings. There usually is no need to fix that part of the intrument sound. But on instruments where a regrad is no option trying differnt strings may be useful. And I understand that many dealers may need that extra variable to play with.

We'll see at some point i'll might need to go down that road.
"If the facts don't fit the theory, change the facts". A Einstein




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users