Case example 1: The clarinet and the WX7 MIDI wind controller.

As an example of this typology, let's first consider a "first order" model of a clarinet, where one gesture accounts for one musical function. We will then show the limitations of such approach and discuss some questions related to the real instrument's behaviour :

• Breath pressure - exciter gesture.
• Lip pressure - parametric modulation gesture.
• Key value (fingering) - selection gesture.

This description applies to a clarinet-like electronic controller such as Yamaha's WX7 MIDI wind controller, where these three gestures are sensed by different transducers and the controller outputs three independent MIDI data streams.

Let's now analyse a real clarinet:

• Breath pressure is mainly an exciter gesture, sustaining the air column vibration. 3
• Lip pressure (embouchure) is mainly a parametric modulation gesture, since it affects the instrument's pitch (frequency vibrato). 4
• Fingering is a mostly selection gesture, where the performer chooses specific keys for selecting a pitch (note). 5
• Changing the size of the instruments tube is a structural modulation gesture - as done in an orchestra to tune the clarinet before performance). 6

In an acoustic single reed instrument, it is well known that the air flow through the reed is dependent on the difference between the pressure inside the mouth and the pressure inside the mouthpiece, for a certain embouchure value [13]. In the case of tight embouchure, even a great variation of breath pressure will not produce a comparable timbral nor loudness variation. What will probably happen in the case of strong blowing is the effect of clamping the reed (and therefore stop sound production). In order to produce a large excursion on timbral and loudness values, one should keep embouchure values low. In consequence, it affects the way the exciter gesture works.Breath pressure (considered as a result from the difference in pressure described above, in the case of a MIDI wind controller) is thus related to a particular embouchure value. These two gestures are dependent in a reed instrument.

One can conclude that for a real instrument, it is not easy to classify gestures according to independent functions, since they may be interrelated. Nevertheless, this gestural typology is useful to describe the gestures sensed by a controller such as the WX7, since it can be viewed as a model of an acoustic instrument. It seems not possible to separate gestures in different independent classes for all existing instruments.

The importance of this typology is to provide general guidelines for someone designing a gestural interface, mostly regarding the presence of different types of feedback related to different gestures. As Cadoz points out: "In particular, in the case of exciter functions a mechanical retroactive from the transducer to the operator was required to enable the instrumentalist to receive the tactile information from the object he was manipulating". [10]

Wed Feb 10 10:07:20 MET 1999