3.2 Analyzing conducting movements during performance
3.2.2 Interview with conductors
The interaction between the conductor and the orchestra is highly complex. When we approach the analysis of real conductors during performance we do not want to model all its complexity, but to learn specific insights to later be used in interactive conducting applications. In this context, we carried out an interview with professional conductors and their students, in an educational context, to get their feedback about how to get the most from analyzing a real performance. We made participants discuss about the following questions:
• What actions of the conductors have a causal and measurable effect on the resulting performance by the orchestra?
• Are there useful insights to learn from actual performances to apply in DMIs based on the conductor metaphor for non-conductors?
Seven participants (all male) attended the interview, which took place in ESMUC. Three of them were professional conductors who teach the Bachelor of Music Degree in the Orchestra Conducting speciality, as well as the Master Degree in Advanced Conducting. The other four were students of this Master program.
3.2 Analyzing conducting movements during performance The interview was organized by asking these two questions to the participants followed by open discussion. We started the meeting by showing the test recordings we already had from rehearsals to explain the kind of recordings we were planning to make and the sort of descriptors that could be extracted from motion. The session was recorded for later analysis of verbal responses. Below we detail the conclusions of such analysis.
Causality and measurability of conductors actions during performance
Conclusion 1 Causal relationships and correlations between conductors gestures and resulting performance are not always present. Participants identified a number of reasons for this:
1. Conductors sometimes decide not to make any actions at specific moments if not required, specially with experienced orchestras.
• “You can get great results performing very few gestures. It is sometimes better
to leave the orchestra play if it is a good one and they are performing the way you want.” [Teacher 1]
• “I saw Bernstein live once. He made a gesture for the orchestra to start
and nothing else. When we feel the orchestra gets to a certain feel, we stop conducting and the orchestra sounds better. Not to conduct is also to conduct”.
[Teacher 2]
2. As suggested by one participant, some gestures contrast to the music in order to express a desired change in the performance. This means that the observed gesture could be seen as contradictory with the concurrent outcome.
• “You might find that an orchestra is playing an exaggerated staccato and the
conductor performs a legato gesture only to correct the performance. The gesture does not always have to be directly related to the sound, he can just be making corrections”. [Teacher 2]
3. The indication of some intention might just appear at the beginning of a phrase and disappear for the rest of it if the orchestra plays as intended.
• “You cannot expect, for example, to see that the conductor makes forte ges-
tures all the time that the music is forte. If you want to indicate forte you just indicate it at the beginning of the phrase, there is no need to be constantly communicating it.”. [Teacher 3]
Conclusion 2 The personal style of the conductor and the experience of the orchestra affects the way in which the interaction occurs.
• “This is all very relative. Even with professionals, every conductor has his own
style.” [Teacher 1]
• “It is not the same to conduct a young orchestra or a professional one, which is
able to immediately react to your gestures.” [Teacher 2]
Insights to be learned for virtual conducting applications
Conclusion 3 Whatever a user will do with an interactive application is not conducting but something else. In this sense, the analysis of the conductor movements during performance has to focus on the specific set of tasks that the user will have in the application (e.g. controlling tempo). One participant (Student 4), who had used an existing interactive conducting application7, suggested that it could actually benefit
from a better knowledge of real conducting gestures in terms of becoming more realistic. • “Whatever a non-conductor will do in front of an application will not ever be
conducting: it is a game. Here we have students who spend six years studying and even so they do not get enough information.” [Teacher 2]
• “Conducting is just something too complicated to be replicated in an application.
As we mentioned, it really depends on the orchestra.” [Student 2]
• “Conducting is not just communicating the pulse. This is something we realize now
as students. It is so complex that I do not think you can automatically measure all its complexity.” [Student 1]
• “There is this application in Vienna where you can conduct an orchestra. The
funny thing is that if you use actual conducting gestures it does not work properly. The gestures you have to use are very hieratic. I guess that could be improved if you actually look at what conductors do”. [Student 4]
Conclusion 4 There are mainly three aspects of conducting movements to observe on an automatic or semiautomatic analysis: synchronization (beat induction), dynamics and articulation.
• “Generally speaking, it is true that if you do a gesture like this [waves his hands
strongly] it is forte and this other way [waves his hands softly] it is piano.” [Teacher
1]
• “I think there are a number of things you could see. For example, legato and
staccato gestures are clearly different and you most likely can see that reflected 7The application this participant referred to is the Virtual Conductor (Borchers et al.,2002), to which
3.2 Analyzing conducting movements during performance
on automatic descriptors. However, you must be careful when relating that to the resulting audio.” [Teacher 3]
• “There are orchestras that follow exactly your gesture and others that take a while
to react. You might be able to measure this.” [Teacher 2]
Summary
Even though at the beginning of the interview participants seemed quite skeptic about any kind of computational analysis of conductors movements during performance, they ended up identifying some elements that could be subject to this kind of study. How- ever, always having in mind the considerations that derive from the aforementioned conclusions:
• We can encounter parts where there is no direct correlation between the conductor’s movement and the resulting performance.
• The conclusions of the analysis of a conductor with an orchestra might not be applicable to other conductors or orchestras.
• Any analysis trying to get insights for the design of virtual conducting applications has to be focused on the specific tasks that users will have. Good candidates in this sense are synchronization (beat induction), dynamics and articulation.