Why is it So Easy to Identify a Human Voice but So Hard to Identify a Violin?

Numerous listening tests have shown that even expert listeners have trouble telling nominally very different violins apart.

By Joseph Curtin | From the November-December 2021 issue of Strings magazine

Players and makers often talk about differences in tonal character between Stradivari and Guarneri del Gesù violins. Here is Yehudi Menuhin on the subject, from his autobiography Unfinished Journey: “[Stradivari] made brilliant, burnished sound that conveys, for me at any rate, moral notions of loftiness. One must rise to a Strad before it will speak from its craftsman’s soul. It spurns the man who lets his hand exert too much pressure or his finger fall ever so slightly wide of its mark.” By contrast “…the Guarnerius, whose earthier voice belies the fact that it is often slightly smaller than most Strads, sings through its pores and sings de profundis. One need not rise above oneself, for it appeals to the natural man. Although Strads have been the dominant instrument of my life, at regular intervals I have played Guarneri; finding the first gold while the second brings to mind the red of Sainte-Chapelle stained glass.”

Numerous listening tests have shown that even expert listeners have trouble telling nominally very different violins apart.

Menuhin was a great violinist, an eloquent writer, and a highly cultivated man, so it is hard to imagine anyone better qualified to speak on this subject. It is not clear, however, how obvious the differences he describes might be to the rest of us. Should any interested listener expect to hear them by comparing recordings of Strad players like Menuhin, Nathan Milstein, and David Oistrakh with those of Guarneri players like Jascha Heifetz, Isaac Stern, and Pinchas Zukerman? Or can the differences be heard only by expert listeners, or by the players themselves?

As it happens, numerous listening tests have shown that even expert listeners have trouble telling nominally very different violins apart. Take for example a blind test organized by BBC radio in the mid-1970s. Violinist Manoug Parikian stood behind a screen and played the opening of the Bruch Concerto and an excerpt from Bach’s Chaconne on each of four violins: a Stradivari, a Guarneri del Gesù, a Vuillaume, and a Ronald Praill that was barely a year old. Stern, Zukerman, and the British violin expert and dealer Charles Beare were each asked to guess which instrument was which. Before giving their answers, the panel spent some time pointing out the test’s deficiencies: They felt the excerpts were too short and too limited in tonal possibilities; that there was no chance to revisit each instrument for extended comparisons; that the studio represented only one of many possible listening environments; and so forth.

In the end, no one correctly identified all the violins. Stern and Beare did best, getting two out of four correct. The Praill violin was mistaken for both the Strad and the Guarneri.

It goes without saying that a healthy skepticism should be reserved for results based on such a small sample size. While the BBC test was not undertaken with sufficient rigor for the results to stand as scientific evidence, this does not mean there is nothing to be learned from it. And yet, fairly or unfairly, listening tests have had little apparent effect on prevailing beliefs about violin sound. Rather than speculating about why this may be the case, I would rather consider a more general question: What kinds of sounds are humans good at recognizing and comparing—and do violins make these kinds of sounds?

As a point of reference, imagine three widely familiar voices—those of U.S. presidents Bill Clinton, George W. Bush, and Barack Obama. How clearly can you hear each of them in your mind’s ear? That probably depends on what kind of mind you have, and what kind of ear. Imagine you have been blindfolded and, after hearing each president say the same simple phrase, are asked to identify which voice is which. Would you feel nervous? Probably not. Humans and many other species are adept at recognizing each other’s voices. Bats, for example, can find their offspring among millions of other baby bats, all crying out simultaneously in a lightless cave. 

Not only do we recognize the voices of people we know, we also recognize general qualities in the voices of strangers—an Irish accent, for example, or the effects of a common cold. The violin (or viola or cello) is often said to be the instrument closest to the human voice. Is it therefore reasonable to suppose that the sound of an individual violin would be as distinctive and memorable as that of an individual voice?


Imagine you have been blindfolded and asked to identify the Soil Stradivari from a lineup of five other violins, having heard the same brief passage played on each. Would you feel nervous? You should.

Well, imagine the sound of a Stradivari violin. What do you hear? To be more specific, imagine the sound of a particular Stradivari, the “Soil” of 1714. This was Yehudi Menuhin’s principal instrument and is now played by Itzhak Perlman. Perhaps you can call to mind a passage from one of their recordings or remember how the violin sounded in a live concert. But now imagine you have been blindfolded and asked to identify the Soil from a lineup of five other violins, having heard the same brief passage played on each. Now would you feel nervous?

You should. 

Notwithstanding everything said and written about the differences between various kinds of violins—old, new, Italian, French, German, English—these differences have been surprisingly difficult to pin down in listening tests. This does not mean that they don’t exist. Variations among individual instruments of the same type can be relatively large, while general differences between types may be quite small, meaning you would need to compare a very large number of violins in order to draw meaningful conclusions. It could also be that some kind of specialized perceptual training is required. Consider, for example, that tiny differences between birdcalls may be obvious to ornithologists who have spent hours listening to recordings of them played at quarter-speed.

From all this, we can at the very least conclude that violins are harder to tell apart than presidents. Why is this so? A few years ago, a client of mine from the Pittsburgh Symphony Orchestra came in for an adjustment and asked about a prototype electric violin that was sitting on a cabinet. The instrument, I explained, was designed to produce the sound of the bowed strings, uncolored by the resonances of a normal violin body. Going to my computer, I played an excerpt from the Tchaikovsky concerto that was recorded with this instrument. If you’ve ever listened to the unmodified sound of a solidbody electric violin, then you know its buzzing, colorless tone. And yet a few seconds into the excerpt, my client said, “Is that Ilya Kaler?” When I asked how on earth she knew, she answered immediately: “The sound!”

The sound? It barely sounded like a violin, let alone Ilya Kaler’s instrument. Kaler had been guest concertmaster with her orchestra the previous year. Clearly, she had recognized something in his sound that was quite independent of his violin. She had recognized, presumably, the collection of technical and musical inflections that add up to a player’s “voice.”

A great violinist once told me that in the old days, you could immediately recognize artists like Heifetz, Fritz Kreisler, Mischa Elman, and Menuhin, but that these days violinists all sound the same. This is a complaint frequently lodged by players and critics. Out of curiosity I asked him if he thought he could, while blindfolded, tell which of his students was playing at a master class. “Yes, of course,” he said. It is tempting to conclude that violinists are better at telling each other apart than they sometimes pretend to be, but in fairness, I am taking the complaint too literally.


What he meant was that, while there are many accomplished violinists in the world today, there are few distinctly original voices.

If we easily recognize speaking voices, and perhaps less easily the musical voices of individual violinists, why is it so difficult to tell violins apart, at least in blind listening tests? Over the past decade, these have shown that experienced violinists and listeners are unable to distinguish Old Italianviolins from new ones at better than chance levels—not by playing them, not by listening to them. If there is indeed a distinct Old Italian sound, why isn’t it as immediately recognizable as, for example, an Italian accent?

An obvious way in which the voice differs from the violin is that “player” and “instrument” are one and the same. You can’t take Clinton’s vocal apparatus and ask Obama to make a speech with it. On the other hand, a gifted mimic can use his own vocal apparatus to imitate both presidents well enough that you would at the very least know which was which. This suggests we recognize each other’s voices more by the way we use them than by their innate tonal characteristics. Were Pinchas Zukerman to play Itzhak Perlman’s violin, most people would agree that he would sound pretty much as he always does. While it is possible to imagine Zukerman using his own Guarneri del Gesù to imitate Perlman’s playing, it is harder to imagine him using it to imitate Perlman’s Stradivari. What would that even mean? All this suggests that the recognizable attributes of violin sound are more related to how the player uses the instrument than to the instrument’s innate tonal characteristics.

There are some physical reasons for this. Most musical instruments can be looked at in terms of two separate systems, an oscillator and a resonator. The oscillator provides a signal that varies in pitch and amplitude rather than timbre. This signal is then “colored” by the resonator. The bowed string and the vocal cords both act as oscillators, while the violin body and the various resonating chambers in the vocal apparatus (lungs, throat, sinuses, mouth) act as resonant amplifiers. 

There is, however, an enormous difference between the vocal tract and the violin body. 


People can freely modify the placement of their vocal resonances, and do so with every change of vowel. A violin is made with wood, so its resonances, or normal modes, are relatively fixed. In this respect a violin more resembles a room than a voice. A room’s acoustical characteristics are also determined by its modes. These can be modified only by physically changing the room—by rearranging the furniture, for example, or removing a carpet or turning up the heat. Similarly, violin modes can be modified by adjusting the setup or changing the fittings. What is important is the relative stability of the violin’s resonant structure while it is being played.

It happens that our perceptions are highly sensitive to changes but tend to ignore things that remain constant. We notice the sound of a refrigerator mainly when it turns on or off. This is also true for complex acoustical constants, such as room acoustics. If you enter a highly resonant passageway while speaking with a friend, the conversation may suddenly become unintelligible. Very soon, however, the brain manages to separate the fixed acoustics of the space from the continually changing signals that make up spoken language. Conversation again becomes possible. 

It is reasonable to assume that when listening to a violin, our auditory system focuses less on the unchanging acoustical characteristics of the violin body than on the continuously changing signal produced by the violinist—that is, the music. If a player switches violins, the difference may at first be noticeable, even dramatic. But the ear soon adjusts to the new acoustical constant and focuses again on what the player is doing. It is for this reason that very short excerpts are used in listening tests.

Just as the ear adapts itself to the acoustical environment, violinists adapt themselves to individual instruments. Great players find ways to project their own distinctive voice using almost any violin. There are limits, however, to what both players and listeners can accommodate. Not all rooms make good concert halls; not all violins make effective musical tools. Violinists may not be able to identify violins by maker or age, but they readily separate instruments they love to play from those they do not. Understanding the physical characteristics that make great violins (both new and old) so intensely appealing to players is an exciting facet of current research. 

Joseph Curtin is a Michigan-based violin maker, researcher, writer, and 2005 MacArthur Fellow. He is co-founder and co-director of the VSA Oberlin Acoustics Workshop, and has co-written three papers on blind-testing violins, published in the periodical of the National Academy of Sciences.