Fingerprinting

[Preston]

To recap:

In this thread we are examining the question: What are the minimum component parts necessary to define a model of a musical fingerprint?

I listed 7 elements suggested by various people in other threads. These were: (i) Intervals, (ii) Harmony/Instrumentation, (iii) Melody/Note progression, (iv) Rhythm, (v) Register, (vi) Dynamics, and (vii) Keys.

To date we have eliminated the seventh: keys.

I would like now to turn to numbers 5 and 6: Register and Dynamics.

I will define these terms as follows:

Register: the range, or characteristic part of that range, of an instrument or the human voice.
Dynamics: (i) indications of loudness/softness, (ii) indications of style such as staccato/legato, and (iii) indications of velocity or velocity change such as lento/presto, accelerando/ritardando.

The question then becomes:

Are either or both of these necessary components in a model of musical fingerprinting?

Or, to put it another way, if they were removed (or ignored, or deliberately changed) when performing any particular score, would such a removal/change alter, weaken, or destroy completely, any musical fingerprint in that score?

Megan, in particular, feels that they are an important part of fingerprinting.

Well, I can recognize Vivaldi by the very bright high register, with a very clear and blold dynamics shape. There's not much room for shadow or a shading of the keys in Vivaldi's musical world.
Beethoven, tends to use notes or key in the middle or lower register and this gives him the flexibility to move which everway he wants. (Megan 9th April)

I beg to differ a little from this view. Except where the changes made to (i) the dynamic markings (exceptionally slow where presto was marked, for example) or (ii) the intended register (playing the high trumpet on the double bass, for example) are wilfully different, I suggest that removing the Register or Dynamics as components in our proposed model would make little or no difference compared with other elements yet to be discussed.

If that is agreed – and I can imagine that not everyone will agree – we are now reduced to 4 possible elements in our model: (i) Intervals, (ii) Harmony/Instrumentation, (iii) Melody/Note progression, and (iv) Rhythm.

Euan

I am a little confused. Euan, am I right in saying that you are wondering exactly what makes us say that sounds like Beethoven's music? I imagine that is an extremely complex question? If so, you would be asking about the precise musical theory behind the "sound" - right? It is too complex for me, by all means - but it is a most interesting question. If so, you are talking about some serious analysis!

[Preston]

Yes there is a superficial similarity to their music coming as it does not only from the same era but the same culture. However the more familiar you become with their music the differences become far more apparent than the similarities. Mozart is a more naturally melodic composer than Haydn and he also tends to have more chromaticism in his music - 'The Dissonance' quartet opening is a good example, the development sections of symphony no.40 (finale) and development section of symphony 41 (1st movement) plus that marvelous late gigue - none of these could possibly have been written by Haydn whose music is more diatonic. Haydn though is generally more experimental - he uses some unusual keys for the time such as C# minor and he often has these contrasts of key between movements within the same work. I also think there is a greater wit in his music than in Mozart - something Beethoven definitely exploited - some examples are the slowing down of the music at the end of symphony no.98 before a sudden rush to the end or the sudden loud chords in symphony no.94 slow movement.
So summing up I'd say that Mozart is really an operatic composer and Haydn more naturally symphonic.

Another weird coupling of composers that is often made is Bruckner and Mahler who are completely different from each other.

Thank you Peter for your post. I do believe that it is more of a superficial comparison also.

What is this late gigue? I am curious to hear it.

[Chris]

What is this late gigue? I am curious to hear it.

Kleine Gigue in G for Keyboard, K574:

http://www.youtube.com/watch?v=hHE4KbY5Uo8

[Peter]

Thank you Peter for your post. I do believe that it is more of a superficial comparison also.

What is this late gigue? I am curious to hear it.

Hardly typical of Mozart as the Chromaticism is extreme, but still somehow discernible as such. I see Chris just pipped me to it!

http://www.youtube.com/watch?v=BGsCE...eature=related

[Preston]

Thank you Chris and Peter. I have never heard that piece, so one of my first thoughts was - is this Mozart?! It does sound different from anything, I have heard written by Mozart.

[Euan Mackinnon]

Peter, Preston

Thank you for your recent comments.

I apologise for the delay in replying but I am on holiday in La France Profonde at the moment and wifi is not that easy to come by . I will reply in more detail when I return to the UK.

Euan

[Megan]

Peter, Preston

Thank you for your recent comments.

I apologise for the delay in replying but I am on holiday in La France Profonde at the moment and wifi is not that easy to come by . I will reply in more detail when I return to the UK.

Euan

Euan, at first I read, that 'wife' is not that easy to come by.
my eyesight gets a bit squiffy reading on the pc screen.
Hope you had a wonderful time.

.

[Euan Mackinnon]

Back from my holiday I would like to respond to some comments before taking the general argument further in a subsequent post.

Preston (13th April)
I am a little confused. Euan, am I right in saying that you are wondering exactly what makes us say that sounds like Beethoven's music? I imagine that is an extremely complex question? If so, you would be asking about the precise musical theory behind the "sound" - right? It is too complex for me, by all means - but it is a most interesting question. If so, you are talking about some serious analysis!

I am sorry if I have confused. Perhaps it would help if I summarise what I am trying to do:

1. Many here (and elsewhere) claim they can listen to a piece of music and tell - even be certain - who wrote it. In short, (at least some) composers (unconsciously?) embed in their music a musical fingerprint. (Incidentally, I don’t dispute their claim.)

2. We must hear something in (say) Beethoven’s music that corresponds to (t)his fingerprint.

3. The music starts from the score so the fingerprint must lie in that.

4. I am exploring what set of elements in any score by any composer is required (i.e. has to be extracted) to identify the fingerprint embedded in that score.

5. In short, I am trying to suggest a model into which these elements are ‘plugged’ and which then predicts (i.e. identifies) the composer.

In my view, there are four distinct stages in producing such a model and we are still on the first stage here, in this thread. These stages are:

A. Reduce the musical elements down to the essential: the ‘necessary and sufficient’ variables.

This is what I have been trying to do so far and have reached the stage where we have yet to consider (i) Intervals, (ii) Harmony/Instrumentation, (iii) Melody/Note progression, and (iv) Rhythm.

Peter (12th April) disagrees; he would retain dynamics and register.
No you can't eliminate [dynamics] - sudden Sf marks, adbrupt changes from Forte to Piano are part of the style of C.P.E.Bach and Beethoven learnt much from him as these characteristics feature a lot in his music. Then what about Signor Crescendo himself - Rossini? As to register, a wide spacing between the hands becomes a part of Beethoven's late piano writing style.

B. Take undisputed works by each composer in turn and run a series of processes on those works to (try to) establish the interrelation between these remaining (‘necessary and sufficient’) variables for that composer.

Of course, we may well expect the interrelationships for composer X to be different from those for composer Y. So, the next stage is to …

C. Distil a common model out from the set of composer-specific interrelationships.

D. Finally, test the model.

I hope that helps to clarify and simplify what I am suggesting in this thread.

One final point: as any scientist, engineer, economist, etc will say, a model is just that: a model. It is not the reality. The reality for us in this thread is the actual music (the score), music in which most of us feel we can hear ‘something’ that says to us ‘Beethoven’ (or ‘Mozart’ or ‘Vivaldi’ etc).

Thus, as for the scientist, engineer, economist, etc, our model will have weaknesses as well as strengths. Furthermore, it is likely to end up being a probabilistic (rather than deterministic) model which brings us back to Preston’s comment above. Still, nothing ventured, nothing gained.

Euan

[Sorrano]

Welcome back, Euan! Another thing to consider is performance practice for a given composer. For example we are taught to play Mozart (keyboard) in one way and Beethoven in a different way. Pedal interpreting, etc. plays roles, too, and I think this concept extends into all genres and instrumentation.

[Quijote]

An excellent point by Sorrano, which answers (in part) Euan's point 3 above ("The music starts from the score so the fingerprint must lie in that."). This is debatable, even highly so. To reinforce Sorrano's posit, there is much that is not in the score; ergo, we must be circumspect here.
I will try to answer Euan's other points later (sorry, holidays, kids, lack of time, blah blah ...).

[Quijote]

Euan, great topic, by the way. Let's follow it as far as we can.

[Euan Mackinnon]

Sorrano (22nd April)
Another thing to consider is performance practice for a given composer. For example we are taught to play Mozart (keyboard) in one way and Beethoven in a different way. Pedal interpreting, etc. plays roles, too, and I think this concept extends into all genres and instrumentation.

Philip (22nd April)
An excellent point by Sorrano, which answers (in part) Euan's point 3 above ("The music starts from the score so the fingerprint must lie in that."). This is debatable, even highly so. To reinforce Sorrano's posit, there is much that is not in the score; ergo, we must be circumspect here.

Philip (22nd April) – New Beethoven’s Fifth
[…] each performance offers (hopefully) new insights, and is not all in the score (Euan take note, please).

Sorrano, Philip

Thank you for your interesting comments and proposing that there is another factor I should take into account when trying to model fingerprinting, namely performance.

I have a problem with this. Logically, it seems to me that there are three possibilities:

1. The score contains necessary, but not sufficient, information to model the fingerprint: performance provides the remaining and necessary element.]

2. The score contains all the required information but there are in fact multiple fingerprints per composer and the performance reveals one of these.

One of several ways of looking at this would be as follows: at any one time there is only one fingerprint but that fingerprint ‘evolves’ as new performance practices take over.

3. Philip and Sorrano are not correct: the score does contain all the necessary and sufficient information required to model a fingerprint.

This is not to say, of course, that the performance reveals nothing new – I have no doubt that it does so – but rather that the fingerprint is independent of the performance (just as it is independent of key - a point we discussed earlier in the thread).

Nor is it to say that one performance cannot ‘bring out’ (i.e. reveal) the fingerprint more clearly than another.

I am interested to hear how Sorrano and Philip (and, hopefully, others) develop their argument. Until then I shall retain the view that possibility (3) is the case: all the necessary and sufficient information (to provide a model of fingerprinting) is in the score. At best variations in performance merely reveal (or obscure) that fingerprint.

Euan

[Peter]

I agree that the fingerprint is independent of the performance - a musician should be able to read a score without performance and still have a pretty good idea of the composer. Problems can arise in identification of some works say perhaps of Haydn and Mozart where they are not at their best, some of the earlier works for example where the style or fingerprint isn't fully evolved.

[Sorrano]

However, the performance is not independent of the fingerprint and can provide clues.

[Sorrano]

As an example of my above post, while playing through a piece by Liszt I considered the pedaling technique used as compared with Chopin. There are a lot of similarities, I think between the composers, but the technique is much different. The overall effect that each was trying to achieve is different. These kinds of things are not expressly integrated in the scores, but through the scores we can get an idea of what messages are being conveyed. However, the actual performance will bring these nuances out that simply cannot be notated via standard notation, i.e., a pause here, a brief accelerando, all created without notational indication to establish the personality of the composer within the work.