A SOFTWARE TOOL
FOR
THE DESCRIPTION
OF
(MUSICAL)PHRASE INTERPRETATION STYLES

In what follows we describe an ongoing project for the development of a computer music tool (in HASKELL), centered on the notion of musical phrase interpretation.
In it (multi-level-)phrase-structured scores can be represented, and it is possible to specify the dynamic, timbral and timing envelopes that the user wish have to govern the progression of the phrases.
Moreover, a number of general rules are implemented among those that are sometimes applied (wrongly or not; it is not our task to judge) by performers in their interpretations. Such rules modify the dynamic, timbral and timing envelopes depending on particular aspects of the phrases. The effects of these rule on the final performance can be disabled or, when enabled, parametrized.
In this tool it is then possible to listen at the overall effect of using particular dynamic, timbral and timing envelopes and of using (or not) particular rules with particular parameters. It will be up to the listener to judge which choices make the music sound "natural", deciding by her/himself whether it is possible even to think about a "general" formal description of music interpretation styles.

For the time being the tool under development takes into account only one melodic line.

Motivations and Brief Description of the Tool under Development

Is it reasonable to talk about "style" in music performance? be it the general performing style of the music of a certain period (classical, romantic,..), the style of an orchestra or, even more particular, the style of a performer?
If so, what is style?
Instead of trying to give a precise formalization of what style is, let us think now about what makes us say that some music we are listening to is being played according to a certain style.
As a matter of fact there is no doubt that we can often know a style even when listening to the performance of a never heard before piece of music. Sometimes good musicians have also the ability to recognize that one player is playing according to the style of some other player or that one orchestra is trying (maybe in vain) to mimic some other orchestra. In general we know a performing style when we recognize some particular patterns in a performance, that is we notice that some elements of the score are expressed in a similar way.
Therefore any reasonable discussion about style need to have as a starting point at least the assumption that scores are looked at by players as structured entities, that is a player (consciously or unconsciously) when reading a score, "decompose" it into "elements", and that the performance of such elements is influenced by particular rules, those that produce the "recognizable" patterns in the hear of the listener.
One more very general and quite reasonable assumption has to be made about such not better specified "elements" and "rules": when such rules are applied on the elements of score what they affect is the dynamic, timbre and (slightly) the duration of the sounds produced by the player (of course, according to the instrument(s) played, timbre or dynamic cannot be affected at all).
The one presented above are very general and vague assumptions, but we believe they are also the only ones which are not arguable when we wish to carry on a discussion about performing styles.
From this starting point on the discussion can begin, but we do not intend to start any discussion.
Our intention is, instead, to offer a tool that could be helpful when discussing or debating about performing styles.
Let us explain how with an example.
When a renowned musician like Antony Pay, talking about classical music performance, vigorously keeps on stressing that "A classical phrase must begin!", this naturally implies that for him, playing according to a good classical style amounts also to follow a rule stating that the beginning of phrases must be made clear.
Even if we personally (and completely) agree with such a statement, there is nothing that could help some arguing going on.
One could then support Pay's statement is several ways. One of the possible ways is no doubt the following one: Let us first word the above rule more formally by proposing a dynamic envelope and a timbric envelope that are in general to govern the progression of a classical phrase, and in which the beginning of a phrase is "stressed".
Then let us try to implement this rule in a software tool (ours, in our intentions).
After that we could consider some classical music and make the software tool play it following the rules implemented.
Now, no argument to support Pay's statement could be better than the fact that what is produced sounds natural to our ears.
One could then also go on and experiment with different ways of stressing the beginning of a phrase, that is providing different envelopes or focusing on the dynamic or timbric one alone.

So we are developing our tool in such a way the user can "type" a score (a single melodic line to begin with) and then decompose it (by hand) into several phrases.
These phrases, in turn, can be decomposed, if we wish, in subphrases down to any desired level of nesting.
Then we can form the style according to which the tool will have to perform the score.
A style will be considered to be decomposed into a dynamic, a timbric and a duration component. Each of these components of a style can be implemented by the user, by setting several parameters. First of all by giving the general (dynamic, timbric, duration) "shape" of a phrase (i.e. a function representing the (dynamic, timbric, duration) envelope of the phrase) together with several other parameters concerning such a shape and its effect on the performance of the phrase [More details in section xx].
Afterwards the user can choose among a wide bunch of rules. These rules describe different ways the dynamic, timbre or duration of a phrase can be affected by particular aspects of a phrase. Here is a possible example concerning the dynamic, in order to clarify the notion of "rule" and of "parametricity".
Note that one can select the rules to be used, possibly none. In this last case one can experiment with the overall effect in a performance by varying the shape of the phrase, the rule used and the parameters to be set.
Our aim is to develop also a graphical interface of the tool.

Detailed description and HASKELL code of what has been already developed

• The score and its representation
• The phrase structure of a score
• Components of a style:
  • Dynamic
    • Setting the general dynamic envelope of a musical phrase
    • Possible rules
  • Timbre
  • Duration