an blog | AJBlog Central | Contact me | Advertise | Follow me:

Stradivarius wanted violins ‘to sing human vowels’

Here’s some more cutting-edge science for the weekend.

Academics have been analysing the sound of Itzhak Perlman’s Strad to discover similarities with the human singing voice, as delivered by a Metropolitan Opera soprano. With me, so far?



Joseph Nagyvary, professor emeritus in biochemistry at Texas A&M, found only two Italian vowel sounds, “i” and “e”. Others sounded French and English.

His conclusion:  It appears that famous Cremonese instruments produce notes that gravitate toward certain type of vowels, implying that old masters could have used vowel identification as a means of quality assurance.


Read the full paper here.

Related Posts Plugin for WordPress, Blogger...


  1. John Soloninka says:

    Just FYI…Prof Nagyvary is seen as rather iconoclastic in luthier circles. He claims to have found “the secret” of Stadivari tone, ascribing it to certain chemical wood treatment. This has been shown to be false on so many levels, and plays into the “myth” of Cremonese instrument uniqueness. He seems to be ascribing to Stradivari the ability to bend a violin to his will to reproduce a voice-like quality. This seems circumstantial, not deliberate – ie the violin and its vibrating elements are naturally like the human vocal chords and vibrating air columns of the throat and nasal passages. The singing phonemes of italian are more like a legato violin passage than the staccato nature of Cantonese. This is just using a convenient correlation that again plays to the Myth of Stradivari. See the earlier posts about modern violins compared with Cremonese golden period instruments….some modern luthiers can do now what Stradivari and others did then, but there is no pretension that modern luthiers are deliberately mimicking human voices.

  2. Fabio Fabrici says:

    vowel formant identification is used for centuries, by old masters until today, to describe timbre of musical instruments. Nothing secret about it and no revolutionary insight, but maybe in Texas ;). “i” and “e” are the two frequency ranges that opera singers train their voices to come “over the orchestra” in tutti forte. It’s called the “singer formant”. Just normal that violin makers would try to amplify that frequency range that gives the musician projection and presence.

  3. If bad science gets you tenure, go for it. Justanother application of Say’s Law “Bad money drives out good”

    • The researcher is a retired professor working in his own time. Save the bile.

      • Fabio Fabrici says:

        a retired biochemistry professor, doing some pseudo-scientific “research” in musical acoustics, to be precise…
        the described method is flawed, simply because it is not standardized in the way the sound is recorded (distance, microphone type etc.) and also because even if it were standardized, the “close distance” (how close?) used is not representative for real world listening to violin music. Timbre changes drastically with distance, due to air absorption and frequency dependent directivity patterns.

  4. Geoff Miles says:

    It’s interesting to see some research into vocal qualities of string instruments. Some time ago I had the chance to compare the sound of some fine old instruments directly with less valuable ones (no Strads, but an Amati and Guadagnini viola amongst others). As a listener rather than a player I’d say that for me the major characteristic that set the finest instruments apart from the lesser ones was a perception that the instruments “spoke” (and I would indeed say this was a vocal quality) in a way that made the upper harmonics appear to stand forward of the fundamental pitch. I notice a lot of other instruments also exhibit this quality to a greater or lesser extent – I also notice that it’s an aspect of sound that is very difficult (or impossible) to record with conventional equipment. In audio it is possible to replicate something of the effect by very small time manipulations of this harmonic area with respect to the area of fundamental pitch. Tiny changes in these timing (phase) relationships have a large effect on perceived audio quality. I have a (completely untested) theory that rather than resultant frequency response or internal acoustics, it is the way that the various internal resonances of an instrument interact to produce specific timing differences between fundamental pitches and harmonics that make the biggest difference in perceived quality between one instrument and another. It may well be that singers also use this side effect of combined resonances to project, and the old makers also understood these things instinctively. I’d be very interested if anyone knows more about this subject.

    • Fabio Fabrici says:

      IIRC, humans react differently to frequency dependent interaural time/phase shifts. You can read about it in Blauert’s publications about Psychoacoustics. Of course what you describe affects timbre and overall loudness totally. The secret is probably to make the single parts resonate as much as possible “in phase” so not to cause cancellations and comb filtering effects. And to produce that “singer formant” (3-4 kHz) for good presence for a solo instrument. I have repeatedly participated in blind tests of violins, Strads, Amatis, Guarneris, etc. and new master instruments. The Strads often have a “character”, strong resonances, smokey sound, singer formant but not too brilliant.

      I think the effect you describe of high frequency being earlier or later in time vs the fundamental frequencies is more influenced by the player by the bowing technique than by the instrument itself.

      • Geoff Miles says:

        Fabio – I posted this and then lost track of the thread so hadn’t seen your response, apologies! I’m aware of Blauert’s work, and I’ve been trying to follow some more recent hearing psychology research, although I’m not an academic and not affiliated to any academic institution, so finding the time and the access to the necessary information is difficult. I’m not necessarily talking about interaural phase or time differences – although these certainly perform a very important function when it comes to localization and source separation. What interests me is the way in which our hearing (even when listening with one ear) uses resonances (and their associated phase shifts) to allow our brains the processing time to interpret time domain information that would otherwise be very difficult to perceive. I have a conceptual understanding of how this works, which may of course not fit with the real science – I’m trying to figure that out, and ask the right sort of questions of interested scientists. Conventional recordings (through conventional microphones and speakers) do not present audio information to our ears in a manner which makes it possible for us to decode in the same way – and they may also contain significant distortion in the region where the most important information is contained (1-4KHz). Try to equalize a modern violin recording to sound like Kreisler and you won’t get anywhere close – and that’s not just to do with the performance. Strangely (and counter-intuitively for a recording engineer) some very old acoustic gramophone recordings (within a very limited range) actually do a better job, because they unwittingly model some of the resonance and phase characteristics of our own hearing. I have a feeling that the same impression of being “in touch” with the gestural qualities of the music that I perceive in some old recordings (and it is a tactile, vocal quality) must also be part of the secret of the power of these old instruments. As you say – strong resonances, and not necessarily an even frequency response at all, but I would suggest that it is the relationships between the resonances (rather than their actual exact pitches) and the effect this has on the time domain behaviour of an instrument which are the factors that create the “magic” – at least when the brain of a talented musician is brought into the whole equation! I’m sure you’re right that a player can also influence these tiny spectral timing differences through bowing – that would certainly explain the ability of certain musicians to make an instrument sound bigger, richer and more three dimensional.

  5. As a player of both modern and historical string instruments, I think it is funny when someone like Perlman plays a Strad or Amati, and feigns interest in the sounds that it was meant to produce by its maker, in the 17th century.

    He’s playing the instrument in a bowdlerized setup including a modern neck, steel strings, modern bow, and hamming up with russian-romantic-school vibrato. By 2013, so little of what Stradivari made is even left – the neck, scroll, bridge, fingerboard has all been replaced many times since it was made – that Strad would scarcely recognize the sound of it if he were here today.

    The reason I point out this fallacy is that the same players who buy Strads and del Gesus as status symbols sometimes express disdain for the “silly” historically informed movement and then paradoxically, when blind tests are administered, cannot tell the difference between their precious Strads and the fine instruments made today. Not trying to be on a soap box, but it is curious, isn’t it?

    • Fabio Fabrici says:

      It’s curious and it also speaks for the mastership of the old makers, that their instruments even in the bastardized form they are transformed into today still produce the most beautiful sound. Some of them. Not all Strads sound very good. Some sound just bad. And some modern violins sound fantastic too. The biggest factor still is the player of course.

  6. Perlman’s violin sings because it is him who plays it, he plays it with all
    his “neshume”, with all his soul and sensitivity. He obviously loves the tune and is serious about giving it all for this *** story. ***Can’t find a proper adjective to describe the film, and what it did to audiences.
    The music went under your skin, it made so many people cry.

  7. Michael Adams says:

    Interesting irony here: the first picture of Perlman (Above) shows him playing a modern instrument outdoors, nothing resembling a Strad. This picture is undoubtedly from his performance at President Obama’s first inaugural ceremony when it was acknowledged that the performers “lip-synched” to a pre-recorded track due to the cold temps

  8. Thank you for that information Norman, of course being a retired professor is why he could get published!
    This is because:
    a) he has an established name
    b)he has money.

    There are plenty of graduates and graduate students who would love to get on a studentship programme for a Mphil or PhD with neither.

    Now being from that illustrious class of 91 who were dropped into the last recession, I am rather cynical. I was one of those for whom, the option of a masters programme in either performance or musicology, either of which I was perfectly capable of completing was not an option due to lack of the reddies…me, and about half of the rest of my year who were not fortunate enough to be scientists!

    Instead those of us who still have half-a brain-cell left are forced to read bad science peddled by has-beens. For those Young Graduates in a similar position, I jolly well hope they have a load of bile ready, or else this professor might like to fund a fully paid PhD studentship for one of them in his former department to make amends.

  9. In reply to Segnor Fabrici, I use other vowels than ‘just i and i: to overcome an orchestra playing tutta-fortza, for a start in Sopranos, those sounds need to be adjusted above G5 as the naophargegeal tract is not long enough in most women to accommodate the resonance, even the late Birgit Nilson had to accomodate some of those vowels.

    We are all trained that provided we use our correct registers, and keep our soft palette’s raised, all our vowels should be intelligible throughout the range.

    This is why technical exercises are performed across all vowels, and using the whole range paying particular attention to the passagi. I have no greater difficulty being heard over a full symphony orchestra singing a than [y].

    • A slight dislcaimer, if I may – we are not all trained for SPEAKING vowels to be intelligible throughout the range. After a certain pitch (depending on the voice type) the vowels do converge more to an “ou” rather than “i”. The listener, as in telephone conversations, has to fill in the missing information, that is, the text. Most self-respecting composers would not have singers sing unfriendly vowels in the upper registers. With sopranos this is usually a crowning la-da-da-ouh!

    • Fabio Fabrici says:

      That’s two different phenomenon mixed up. One is actually producing actual vowels above a certain range that is not suitable for it. See Cabbagejuice’s comment. It’s pretty much impossible to produce all vowels intelligibly over the whole range, at least for opera singers who have to sing with a certain loudness.

      The other phenomenon is the singer formant, which is only indirectly related to the actual vowel, but rather a frequency “filter” overlayed over the whole sound production to give the voice presence and range. Singers usually lower their larynx and thus extend the resonating space in their singing tract in order to produce the formant.

  10. As a violinist, I would venture to say that this idea has been around for a long time – as in the musical instructions of “parlando,” “cantabile” and “portamento” – and add that classical music has also been heavily influenced by the rhythms (and, presumably, consonants and vowels) of the Italian language for several centuries.

    Many years ago Professor Nagyvary contacted me, asking me to participate in a “scientific” survey on the tonal characteristics of fine violins. He asked me to suspend a “good quality” microphone and stand a number of feet below it, and record myself playing a scale in loud and soft dynamics and then send him the recording for “analysis.” I didn’t bother, as it was clear that the entire procedure was heavily flawed and that the results would be nearly meaningless, despite the aura of “scientific method” in his eventually-published findings.

an ArtsJournal blog