97 replies to this topic

[Bruce Tai]

My new article has just been published in the Savart Journal:

Stradivari violins exhibit formant frequencies resembling vowels produced by females

(The Savart Journal is an open-access online journal dedicated to string instrument research, articles downloadable in PDF)

In collaboration with Chi Mei Museum (Dai-Ting Chung is the curator of violins there), we recorded four Strad, two del Gesu and eight more historical and modern instruments. Although a violin does not apparently sound like the human voice, my analysis found that it shares the basic formant features of the human voice. The Strads stood out in our analysis by producing higher formants that mimic female vowels. Ordinary master violins are closer to male vowels. We propose that the female quality of Stradivari violins may explain their sweet and brilliant sound. And our data show that del Gesu violins also share this feature to some extent. This is the theory that we presented in the paper, but what do you guys think?

From what I have heard, a few violinists can hear vowels on most violin notes. But most do not directly associate violin sounds with specific vowels. And I have never heard people describe that Strads have a more feminine voice than other violins. It will be interesting to hear the personal experience of MN'ers with regard to associating violin tones to human voices.

[Guy Harrison]

My new article has just been published in the Savart Journal:

We propose that the female quality of Stradivari violins may explain their sweet and brilliant sound. And our data show that del Gesu violins also share this feature to some extent. This is the theory that we presented in the paper, but what do you guys think?

From what I have heard, a few violinists can hear vowels on most violin notes. But most do not directly associate violin sounds with specific vowels. And I have never heard people describe that Strads have a more feminine voice than other violins. It will be interesting to hear the personal experience of MN'ers with regard to associating violin tones to human voices.

Congratulations on your article- I'll have to carefully read through it. A while ago Guy Rabut wrote and talked about describing violin sound as vowel sounds. In his example, it was for soundpost adjustments. He was trying to have a common language that violin makers/repairs and musicians could use, rather than the usual words used to describe sound like 'dark' or 'deep' etc. which can mean different things for different people. Anyway, personally I like comparing violin sounds to the human voice, at least as an additional way to describe sound.

[La Folia]

Some time ago Anders Buen assembled sounds of the open G string on many Strads and Guarneris. I wrote that when they were played sequentially, alternating Strads and Guarneris, one could clearly hear the difference. This applied to the first few violins, which Anders said were stereotypical of the two makers. I don't recall whether there was a recognizable visible difference in the spectra.

It's also not clear whether this audible difference extended to the other violins, however, and I believe that's the crux of the question. Some say that on average there is no discernible difference, but this might be testable with Anders' data set.

[carl stross]

The Strads stood out in our analysis by producing higher formants that mimic female vowels. Ordinary master violins are closer to male vowels. We propose that the female quality of Stradivari violins may explain their sweet and brilliant sound. And our data show that del Gesu violins also share this feature to some extent. This is the theory that we presented in the paper, but what do you guys think?

From what I have heard, a few violinists can hear vowels on most violin notes. But most do not directly associate violin sounds with specific vowels. And I have never heard people describe that Strads have a more feminine voice than other violins. It will be interesting to hear the personal experience of MN'ers with regard to associating violin tones to human voices.

I think it is absolutely brilliant. Breath of fresh air and superb constructive approach.

[carlobartolini]

Congratulations Bruce, very interesting concept and lots of dedication and time there, thanks for sharing.

Is it possible to hear some of the recordings? I tried but had a hard time imagining the sounds that you wrote about, my limited brain hit a wall, hearing would make it easier to understand, for someone like me who does not know by heart the different qualities of these instruments. Plus it would be nice to go back and forth between the text and audio files while reading it all again.

[Torbjörn Zethelius]

I've always thought that it was common knowledge (or myth) that violins mimics the human voice. For me, the castrato voice is the model for the violin.

[robertdo]

After all the human voice is produced by 2 cords vibrating inside the throat, being amplified by a cavity, and being modulated by the shape of the mouth/lips/tongue (vowels-consonants). this is quite a close analogy with the string/bridge vibration, amplification by the closed body and "modulated" by the f-holes.

[Wm. Johnston]

I think it is absolutely brilliant. Breath of fresh air and superb constructive approach.

Actually it sounds to me like returning to the early days of violin acoustics research. Have you looked through the compilation of articles that Hutchins put together in the 1970s? Looking at band averaged violin spectra (formant structures) was one of the early approaches taken by researchers. I don't mean that in a disrespectful way, one of the papers in that compilation was very influential on me.

Also I remember reading in the older (non-acoustics) literature about how Amatis had a pure soprano sound while golden period Strads had a mezzo-soprano sound. If I was in my office I would look for the older articles that I remember but I can't access them right now.

[Wm. Johnston]

being amplified by a cavity

I can't help but comment on this one. I think the throat would just be a passive resonator. I'll leave it at that for now since that is not the purpose of this discussion.

[robertdo]

I can't help but comment on this one. I think the throat would just be a passive resonator, an amplifier requires an input of energy.

Yes that is right, I was more talking about what we commonly describe as "avoir du coffre" in french when it comes to opera singers. It's combination of the lung/chess capacity because the opera singers were often caricatured as quite big fellows in this area, even the women. But of course it's not an exact description, and La Callas was not exactly shaped like Monserrat Cabale...

[Wm. Johnston]

From what I have heard, a few violinists can hear vowels on most violin notes. But most do not directly associate violin sounds with specific vowels. And I have never heard people describe that Strads have a more feminine voice than other violins.

Actually, as I said above, in the older (non-acoustics) violin literature Strads were described as mezzo-soprano and the Amati violins were described as soprano which I interpret as more feminine.

It will be interesting to hear the personal experience of MN'ers with regard to associating violin tones to human voices.

Since I'm currently on vacation and traveling I only read through the article quickly. One thing that I didn't notice was the state of the player when the recordings were produced. Was the player blinded in any way to prevent them from knowing which violin they were playing? It seems to me that a player could inadvertently change how they were bowing a violin and change how the sound. If someone is playing what they know to be a multi-million dollar violin aren't they going to put more effort into playing to the best of their ability compared to a $15,000 violin?

Did you study how the player affects the measured formants of a single violin? Did you look into how the measured formant properties change with the bowing position on the strings, bowing speed, and pressure on a single violin? How do they compare with the spread of the different violins that you present?

[carlobartolini]

Was the player blinded in any way to prevent them from knowing which violin they were playing? It seems to me that a player could inadvertently change how they were bowing a violin and change how the sound. If someone is playing what they know to be a multi-million dollar violin aren't they going to put more effort into playing to the best of their ability compared to a $15,000 violin?

IMHO it depends on the player, it's not that difficult to remove yourself from that sort of bias when playing, if one concentrates hard enough and occupy the brain with other more important issues, as repeating force, precision etc.

I believe the instrument itself can be more of a creator of change, unless we are talking of a player with weak emotion control, and they are out there, specially amateurs, but i believe in such a situation the player is pro and can tune himself out of it, at least to a degree that importance is smaller that the instrument itself.

edit - now that I'm thinking about it, perhaps could the blindfolding not work against a player who is not naturally blind? Being that sometimes when we want extreme precision from our playing, the eyes are a powerful instrument.

[Andrew Kochie]

Just finished reading the full text. Bruce - this is amazing stuff and just answered quite a few "gut-feeling" ideas that I had. Going to have to read through this again.

[Wm. Johnston]

The issues that I bring up have as much to do with the repeatability of the experiment as it does with the player biasing the experiment. Also, I'm the type that wants to be sure that the player wasn't biasing the experiment before I try to learn anything from it. These two things are not completely unrelated. For instance, if you could show that the player can't significantly affect the results of these measurements then that would indicate that the violin contributes something to a performance that a violinist can't control. If, however, the violinsist can change the measured properties by more than is seen in this test then it would indicate that they have a stronger influece on the tone than the violin does.

If you are going to use a violinist to test drive violins then it seems to me the first thing that you need to know is how much of an effect they are imparting on the violin. It's possibile that the violinist had the effect of increasing whatever differences there were between the violins but it is also possible that they decreased the differences between the violins.

In addition how do we know that these differences translate between violinists? Were the tests repeated with more than one person playing the instruments? Are Strads always more 'feminine' sounding than other violins or is it due to how this person plays them? Is it possible that a different player would make the other violins sound 'feminine' and the Strads 'hyper-feminine' to the point that they didn't sound good? I don't think anyone knows that answer, today.

There are other things about this paper that I will probably bring up soon but for now I'll leave it at the violinist level here since most people here are violinists.

[James M. Jones]

After all the human voice is produced by 2 cords vibrating inside the throat, being amplified by a cavity, and being modulated by the shape of the mouth/lips/tongue (vowels-consonants). this is quite a close analogy with the string/bridge vibration, amplification by the closed body and "modulated" by the f-holes.

Both rely on a constant energy input (lungs and or bow) to do the job as well...the violin family of instrument is the closest mechanical system we have to the human system of sound creation....
I don't think the throat qualifies as passive resonator....
I learned to associate vowels w/ tones as a method of figuring overtone content,here on MN some time ago,Mr B,I think turned me on to it, and it was not new then....

[robertdo]

Bruce, just a small question about the vowels sounds. one table gives some phonetic correspondence. Do you think there would be some difference (sometimes strong ones) in the way different people from different nationalities hear vowels?

[carl stross]

Actually it sounds to me like returning to the early days of violin acoustics research. Have you looked through the compilation of articles that Hutchins put together in the 1970s? Looking at band averaged violin spectra (formant structures) was one of the early approaches taken by researchers. I don't mean that in a disrespectful way, one of the papers in that compilation was very influential on me.

Also I remember reading in the older (non-acoustics) literature about how Amatis had a pure soprano sound while golden period Strads had a mezzo-soprano sound. If I was in my office I would look for the older articles that I remember but I can't access them right now.

...and the air was fresher then !

It is nothing new in the fact that great violin tone resembles a female voice or sounds like vowels. Everybody and his brother had this figured out 300 years ago and fine points of "solfegium" and harmony were taught with this in mind ( by the time we stopped identifying notes with our finger joints...). Organ registers were designed to resemble human voices ( Vox Humana, Chours, etc) and some instruments are actually more realistic at that than a violin.

Bruce,brilliantly, puts numbers on things and that's what science needs and does. Now we have a way to quantify the resemblance. Great stuff, Bruce !
And no listening needed !

[carl stross]

Both rely on a constant energy input (lungs and or bow) to do the job as well...the violin family of instrument is the closest mechanical system we have to the human system of sound creation....

Nope. That would be the clarinet.

[carlobartolini]

The issues that I bring up have as much to do with the repeatability of the experiment as it does with the player biasing the experiment. Also, I'm the type that wants to be sure that the player wasn't biasing the experiment before I try to learn anything from it. These two things are not completely unrelated. For instance, if you could show that the player can't significantly affect the results of these measurements then that would indicate that the violin contributes something to a performance that a violinist can't control. If, however, the violinsist can change the measured properties by more than is seen in this test then it would indicate that they have a stronger influece on the tone than the violin does. If you are going to use a violinist to test drive violins then it seems to me the first thing that you need to know is how much of an effect they are imparting on the violin. It's possibile that the violinist had the effect of increasing whatever differences there were between the violins but it is also possible that they decreased the differences between the violins. In addition how do we know that these differences translate between violinists? Were the tests repeated with more than one person playing the instruments? Are Strads always more 'feminine' sounding than other violins or is it due to how this person plays them? Is it possible that a different player would make the other violins sound 'feminine' and the Strads 'hyper-feminine' to the point that they didn't sound good? I don't think anyone knows that answer, today. There are other things about this paper that I will probably bring up soon but for now I'll leave it at the violinist level here since most people here are violinists.

Don't disagree with you, but I have confidence in jaded players that have lots of control and that the enchantment of the name does not really make much effect...if that's the case, get another player, pretty easy to find them out there, only an opinion. (unless one of them is a personal friend of Antonio trying to help his buddy )

And no listening needed !

Carl, could you please e-mail me some of these years of listening to all kinds of violins, I could use it....

How does the Viotti-Marie Hall sounds???

[James M. Jones]

Nope. That would be the clarinet.

unless I'm missing something,
the clarinet is a single fixed point reed ...whereas the voice box has double fixed point cords...mechanically closer to what the violin is doing.