Fritz-blind test

[Doc Eastman]

You are pretty right. First huge peak: 422Hz (a sharp G#₄), next: 619Hz (D#₄), the cluster max above 1kHz: 1249Hz (slightly sharp D#₆), then there is a cluster around 2,8kHz (F₇), and the next just above 4kHz (B₇-C₇), basically an overtone or veery high on the e.

There are dips around 1.72kHz (flat A₆), 2,4kHz (D₇-D#₇), 2,55kHz (sharp D#₇ or a quarter note over) and 2,9kHz (F#₇).

I think you mis-typed d#4, as d#5 is 622Hz (at least according to my Wikipedia table...).

[Anders Buen]

I think you mis-typed d#4, as d#5 is 622Hz (at least according to my Wikipedia table...).

Yes, thanks. It is in the middle of the night here, and I need to get some sleep before going to work. :-)

[Doc Eastman]

No this is right.

For those who don't look at audacity FFTs, softer is more negative. So -30dB is louder than -60dB. So violin 1 at -30dB - Strad at -60dB = a point that plots at + 30dB (-30-(-60)= +30)above the zero line on the plot, or louder on the violin 1 (modern) side. Gee I hope I have this right..algebra happened many moons ago.

[Anders Buen]

For those who don't look at audacity FFTs, softer is more negative. So -30dB is louder than -60dB. So violin 1 at -30dB - Strad at -60dB = a point that plots at + 30dB (-30-(-60)= +30)above the zero line on the plot, or louder on the violin 1 (modern) side. Gee I hope I have this right..algebra happened many moons ago.

Yes this is right. Audacity uses 0dB as the highest possible level and reference, so any natural sound will have negative dB's.

[Doc Eastman]

Yes, thanks. It is in the middle of the night here, and I need to get some sleep before going to work. :-)

We have been speculating on the time difference and wondered how many cups of coffee you were working on.

Sleep well!

[JSully]

it obvious that some people are not hearing the same recording, or maybe its just the low quality of the average computer speakers,

While that's certainly true enough,

i have home stereo speakers hooked up to my computer

Remember that you need a good sound card if you're really going to get thorough about these things. Or a higher quality motherboard, at the very least.

[Addie]

I am stretching the speculations quite far now. But I believe the Modern has a higher tuned bridge than the strad.

Can you expand on this (after a night's rest)?

Most of the lit. leans towards lower tuned for more modern instruments. Also a string issue? Zyex are supposed to be good for modern instruments, i.e overly bright/harsh instruments.

OT, have you ever tried Curtin's experiment of wedges in the kidney openings, to stop the in-plane (side-to-side) movement? I found the results startling, but don't have a good enough mic for FFT's.

[DMartin]

Seems that an obvious next step would be to continue testing the top responders to find if there is indeed a population that can discrIminate old and new instruments (or makers even).

I expect there is such a group.

[La Folia]

Those difference charts show the noise just as much as the peaks, so unless you strip that out, it's hard to see anything from them. (EDIT: They can also have a lot of artifacts, as on the steep edge of a peak.) Here are some plots I think you will find easier to view.

As Anders and someone else pointed out, you can easily hear the difference in the first two notes, and on the E string in the middle of the clip. The Strad is just richer there, with a bigger sound, while the other one is a bit nasal and brash to my ear. It's hard to know why the test subjects preferred the violins they preferred.

I think it's odd that most people who posted comments to the article claimed to identify the Strad, while musicians who were there did not. I have no idea what's going on, and I wish I had been there. I also identified the Strad, listening on my computer with headphones, but I'm pretty sure I could be fooled with a different choice of violins.

EDIT: I have spent some more time with the sound files, and I'm not certain that these samples are reproducible or representative. The results vary a lot from note to note, and the relative results between the two violins vary a lot from one section of the music to the next. I don't think both sound samples are played quite the same, and this seems to make a huge difference.

Note that this problem is not due to difficulties in processing the sound samples. The problem, in my view, is that the samples themselves may not be completely representative. Now if the passages were longer and the violinist played the same passage multiple times on each violin, the power to distinguish between the violins would be MUCH better.

[skiingfiddler]

Seems that an obvious next step would be to continue testing the top responders to find if there is indeed a population that can discrIminate old and new instruments (or makers even).

I expect there is such a group.

Here is a Maestronet post, # 90, from Roger Hargrave, dated June 17, 2011, in a thread actonern posted entitled "question for Roger Hargrave." The quote deals, in part, with the possibility of identifying instrument makers from the sound of the instrument:

Begin quote:

I'm sorry but this is just nonsense. No-one in the world can identify instruments by their sound. Many have tried and being found very lacking. Not even that old chestnut about a del Gesu sound being different to a Strad sound is correct. There is no such thing as an acoustic fingerprint. If there were we would all be using it to identify fiddles. In fact it would be helpful if we could just identify schools, but we simply cannot. These Cremonese/Italian sound stories are largely the product of dealer fantasy. There are many fine sounding modern violins and many shit sounding Cremonese violins. Now, it may be possible to hear differences between good and shit, but "one man's meat" etc. It the past there were always good and bad and mediocre makers and that has not changed. Classical violins have many wonderful attributes and if I had the bread I would buy lots and lots of them, (especially the Amati family). But sound wise there are modern alternatives, however, as I indicated the are good and bad and mediocre modern makers as well. This is not to say that Cremonese violins 'generally' do not sound great, they do. This is because everything taken together these makers made the greatest violins of all time. But this superiority is the sum of all their attributes, and also of the many great restorers who have prepared them for the concert stage.

Just another comment or two in favor of common sense, even if there are two themes mix up together here.

End quote.

Here's actonern's whole thread.

[Addie]

This is getting quite interesting. There was a topic some months ago where Strad and del Gesù clips were commented on. I think I was the only one that preferred the Strad?

So we get down to personal perception, and popular perception. Nobody wants to admit the modern was better here, but in the other topic the cult of del Gesù prevailed.

I like the Strad sound, and I like it for a particular reason. I don't hear that sound in clips of GdG's. Call it "pathos," especially in the lower end. I'm not saying all Strads have it, but it is a tonal quality I associate with Strads, and like to hear in whatever brand of fiddle. I'm not saying I can always pick out a Strad, but I can pick out the Straddy tone quality.

I'm not as sensitive to the differences in the high end that people are discussing. I find this fascinating. So, in general, do you like Strads, and what is it about them that appeals or repels? Do you hear the high end difference? The Low end richness? Be honest, this is science!

Addie, who thinks the Cannone needs a bit of oil to stop that squeak.

[Carl Stross]

Here is a Maestronet post, # 90, from Roger Hargrave, dated June 17, 2011, in a thread actonern posted entitled "question for Roger Hargrave." The quote deals, in part, with the possibility of identifying instrument makers from the sound of the instrument:

Begin quote:

I'm sorry but this is just nonsense. No-one in the world can identify instruments by their sound. Many have tried and being found very lacking. Not even that old chestnut about a del Gesu sound being different to a Strad sound is correct. There is no such thing as an acoustic fingerprint. If there were we would all be using it to identify fiddles. In fact it would be helpful if we could just identify schools, but we simply cannot. These Cremonese/Italian sound stories are largely the product of dealer fantasy. There are many fine sounding modern violins and many shit sounding Cremonese violins. Now, it may be possible to hear differences between good and shit, but "one man's meat" etc. It the past there were always good and bad and mediocre makers and that has not changed. Classical violins have many wonderful attributes and if I had the bread I would buy lots and lots of them, (especially the Amati family). But sound wise there are modern alternatives, however, as I indicated the are good and bad and mediocre modern makers as well. This is not to say that Cremonese violins 'generally' do not sound great, they do. This is because everything taken together these makers made the greatest violins of all time. But this superiority is the sum of all their attributes, and also of the many great restorers who have prepared them for the concert stage.

Just another comment or two in favor of common sense, even if there are two themes mix up together here.

End quote.

Here's actonern's whole thread.

Roger is a clever chap and a GREAT maker but the respective post is at best confusing. Quite a few sentences state the obvious and nobody in his right mind would dispute them. As to "No-one in the world can identify instruments by their sound" that simply tells me Roger has never listen to say, a q/tet where the Cremona one sticks out like a sore thumb. And, by the way, I heard modern ones doing that as well.

ALL great 20th cent. players played on Cremona instruments ( long list here...). Is there ANYBODY on this board who would've argued with Menuhin that Khevenhuller is nothing special and not identifiable by sound ???????????????????????????????? Or Soil ? Or Kreisler's DG ? Or Davidov ? Or have the audacity to present the owners with the "modern alternatives" ???

[jacobsaunders]

Roger is a clever chap and a GREAT maker but the respective post is at best confusing. Quite a few sentences state the obvious and nobody in his right mind would dispute them. As to "No-one in the world can identify instruments by their sound" that simply tells me Roger has never listen to say, a q/tet where the Cremona one sticks out like a sore thumb. And, by the way, I heard modern ones doing that as well.

ALL great 20th cent. players played on Cremona instruments ( long list here...). Is there ANYBODY on this board who would've argued with Menuhin that Khevenhuller is nothing special and not identifiable by sound ???????????????????????????????? Or Soil ? Or Kreisler's DG ? Or Davidov ? Or have the audacity to present the owners with the "modern alternatives" ???

I don't find Rogers Quote confusing at all, nor did I notice him suggesting to argue with Menhuin or anybody else.

[Anders Buen]

Begin quote:

I'm sorry but this is just nonsense. No-one in the world can identify instruments by their sound. Many have tried and being found very lacking. Not even that old chestnut about a del Gesu sound being different to a Strad sound is correct. There is no such thing as an acoustic fingerprint. If there were we would all be using it to identify fiddles.

Here's actonern's whole thread.

With all due respect, this is basically BS.

[Anders Buen]

Those difference charts show the noise just as much as the peaks, so unless you strip that out, it's hard to see anything from them. Here are some plots I think you will find easier to view.

There is information in the noise from the fiddles, so stripping the noise away would not be a good idea. Besides, separating those two components takes some really tricky filtering. I can't separate them, but know one who can. He made his filters himself.

How did you filter?

And what is the frequency resolution you are using?

[Anders Buen]

I think it's odd that most people who posted comments to the article claimed to identify the Strad, while musicians who were there did not. I have no idea what's going on, and I wish I had been there. I also identified the Strad, listening on my computer with headphones, but I'm pretty sure I could be fooled with a different choice of violins.

I think the conditions the players were working under was much more complicated than just comparing two fiddles. I do not think they knew how many fiddles were in the test. Here we know the answer, and there will be an overrepresentation of those who got it right to post it and those who did not not to response.

It is always easy to see and understand things in detail when we already know the answer.

[Anders Buen]

Can you expand on this (after a night's rest)?

Most of the lit. leans towards lower tuned for more modern instruments.

It is an intuition based on what I hear and how the spectra compares. I also have a little experience with such high tuned bridges and what they do to the sound. What literature are you referring to?

OT, have you ever tried Curtin's experiment of wedges in the kidney openings, to stop the in-plane (side-to-side) movement? I found the results startling, but don't have a good enough mic for FFT's.

Yes. (That is by the way an 'older trick' than Curtin). I think it is better to use real bridges with no wedges for a valid comparison. My experience on this is very limited, by the way.

[JohnCockburn]

Anyway. What's the big news here? Hasn't this kind of thing been done several times before?

[Michael.N.]

This link tells us a little more about the test. Note that the instruments smell was indeed masked:

http://www.guardian.co.uk/music/2012/jan/02/how-many-notes-violinist-stradivarius

[Torbjörn Zethelius]

With all due respect, this is basically BS.

For once I agree with Anders.

[martin swan]

You certainly can't get much of a useful sense of a fiddle from a close miked mono recording - they both sound awful to me. The player seems to like the Strad a bit better (there are a couple of cheeky slides in that performance), but the differences I hear are pretty typical of the differences I generally observe between a good new violin and a good old one. The old violin has some narrow frequency bands missing, in quite an attractive way - the new violin has everything in spades, including volume at close range, and a slightly heavy honk a few notes up the A string.

But for the amount of attention we lavish on this subject you would expect massive differences. In reality, there's almost no impact on the musical performance, indeed the benefit of the Strad may be a slightly smoother or thinner neck, or a body resonance which gives the player a bit more feedback and is irrelevant to the listener.

I think the experiment is entirely legitimate, it's all been demonstrated before, but there is zero possiblity of this affecting the market for old violins. if it makes a few more professional players open to the idea of using a good new instrument, or at least not thinking they automatically need to spend a minimum of 6 figures to have any credibility as a player, then it gets my vote!

[David Burgess]

With all due respect, this is basically BS.

I don't think so.

I don't think Roger is saying that one can't hear differences, or that one can't learn to identify a specific fiddle, with enough prior exposure. I believe that what he's saying is that with a group of unfamiliar instruments, people can't identify who made them (including Stradivari), or even tell for sure whether they are old or new (given decent new instruments), by the sound alone. Tests such as the one which is the topic of this thread, and the BBC broadcasting test tend to confirm this. In addition to that, people like Roger have their own experiential data base with players and pricey fiddles, and the kinds of things which happen when they are handed a fiddle, not knowing who the maker is. I've done some dinking around with this myself, where a player will swoon over a copy, or an uncommonly bad sounding Strad. It's highly entertaining what can happen when someone believes they are playing or hearing a Strad.

Anyway, this is stuff that's pretty well known to people in the trade who have had a lot of access to these instruments, like Roger. The only thing new here is that this experiment has made it into the public domain. As usual, it will be met with some howls of protest, and excuses about how the experiment wasn't done right. That's normal and quite predictable. Some kids cry, or refuse to believe it when they find out that Santa Claus isn't real.

Unlike the bulk of experiments which have made it into the public domain, this one also addresses one common excuse:

"Maybe the audience couldn't tell (or the audience was uneducated), but the real advantage is what the instrument does for the player."

In this case, apparently it was the players who couldn't tell, or got it wrong.

From the article:

"And this experiment asked seasoned violin players, not listeners, to choose."

(Anders, this isn't directed at you. I just used your post as a jumping off point)

[Anders Buen]

This link tells us a little more about the test. Note that the instruments smell was indeed masked:

http://www.guardian.co.uk/music/2012/jan/02/how-many-notes-violinist-stradivarius

Reading this I understand that the procedure they used has been quite similar to the tests they did at Oberlin in 2010. (The instruments in that test were among those the acoustics workshop participants brought. A few of the makers also participated in the test. After the test we got a list of all spontaneous comments the testers gave in an anonymous form.).

Reading this article I see that my comment that the players did not know the exact number of instruments was not correct.

[Anders Buen]

I don't think so.

I don't think Roger is saying that one can't hear differences, or that one can't learn to identify a specific fiddle. I believe that what he's saying is that with a group of unfamiliar instruments, people can't identify who made it (including Stradivari), or even tell for sure whether it's old or new, by the sound alone. Tests such as the one which is the topic of this thread, and the BBC broadcasting test tend to confirm this. In addition to that, people like Roger have their own experiential data base with players and pricey fiddles, and the kinds of things which happen when they are handed a fiddle, not knowing who the maker is. I've done some dinking around with this myself, where a player will swoon over a copy, or an uncommonly bad sounding Strad. It's highly entertaining what can happen when someone believes they are playing or hearing a Strad.

Anyway, this is stuff that's pretty well known to people in the trade who have had a lot of access to these instruments, like Roger. The only thing new here is that this experiment has made it into the public domain. As usual, it will be met with some howls of protest, and excuses about how the experiment wasn't done right. That's normal and quite predictable. Some kids cry, or refuse to believe it when they find out that Santa Claus isn't real.

Unlike the bulk of experiments which have made it into the public domain, this one also addresses one common excuse: "Maybe the audience couldn't tell (or the audience was uneducated), but the real advantage is what the instrument does for the player." In this case, apparently it was the players who couldn't tell, or got it wrong.

From the article: "And this experiment asked seasoned violin players, not listeners, to choose."

(Anders, this isn't directed at you. I just used your post as a jumping off point)

Yes, good clarification points, as always. Something like we can't recognize a person from just hearing the persons voice for the first time. If we have heard the voice before, we know. And the likelyhood of nailing a fiddle we have heard before will increase. I nailed three of the fiddles in the 9 instrument set at the 'half double blind' test in Oberlin in 2010. One of them I had played a year ahead, another one had been in my home for about a year and the third I had just tested the day before in the workshop room. The rest of them were unknowns. I thought Joseph Curtins violin was one made by Sam Z.

I think that every instrument does indeed have an individual 'acoustic fingerprint' and basically I think that knowledge is an engineers table, or personel with experimental experience on that, no matter how experienced a qualified maker intuition may be.

[Anders Buen]

There is one aspect to this experiment that I think may play a role. The modern instruments that were included in the listening test were picked out from a set of modern violins by two competent maker and player related pro's.

So then comes the question, did they measure Strads against modern violins or the pro's ability to pick instruments that could match good Strads?

We may argue that the Strads also have been 'hand picked' musical instruments by highly competent testers through the history.