Extended Control Strategies for Woodwind Performance of Electroacoustic Music |
||
Michael F. Zbyszynski Gassmmann Electronic Music Studio |
||
abstractMy experience as a composer, improviser, and performer of music for woodwinds or woodwind-like controller and electronics has shown that many expressive performance gestures are not captured, and the relationship between performer and technology is thereby limited. This is partially due to a lack of continuous controllers, even when playing a MIDI instrument such as the Yamaha WX-7. Ultimately, the MIDI representation of note events is fundamentally at odds with music generated by wind instruments. However, there is still much potential for improvement within the MIDI standard, which can be reached via by adding more sensors to pre-existing instruments, and communicating with a flexible MIDI application such as Max/MSP. This paper will begin by briefly examining some of the control options currently available to woodwind players: foot pedals, MIDI controllers, and pitch/envelope tracking, with the intent of outlining a performance practice for these commonly employed devices. Each offers specific advantages and disadvantages, depending on the musical setting. The rest of the paper describes ongoing research in adding sensors to an acoustic flute and to a Yamaha WX-7. The goal of this work is to broaden the performer's potential for expressive control in a way that feels natural, by starting from traditional performance gestures and placing new controls comfortably within the player's reach. 1. commonly employed devices1.1 foot pedalsThe first electronic controller in a contemporary woodwind performer's repertoire should be the foot switch. Simple, inexpensive, durable and readily available, a foot switch offers the most basic kind of interaction, and should never be overlooked. Why go through the effort of programming a complicated score following environment, if a few cues will suffice? Most performers are used to tapping their feet rhythmically, and can execute pedal cues with a minimum of practice. It is helpful to tape the foot pedal down during the performance; it can be quite distracting to chase an unruly pedal around the stage. The obvious drawback the lack of information that can be communicated. While there are many situations when on or off is enough, a foot switch is not helpful for refined control. The dramatic situation is also limited, although sometimes a timely stomp can be very compelling.  There are two
basic foot pedal designs. My experience has led me to prefer foot switches
that most closely resemble the pedal on an acoustic piano (e.g. figure at
left). These have a longer throw which provides two advantages. First, I can
lightly place my foot on the pedal before an upcoming cue, and be sure of its
location. Second, there is more positive feedback which the pedal is actually
depressed; I can feel the pedal move and then hit the bottom. In either case, these pedals were designed to function in conjunction with a keyboard-style MIDI controller. They usually do not produce MIDI, but plug in to another device that does. This can be an irritation, especially if the performance does not require a MIDI keyboard. To some extent, the portability of a footswitch is compromised by the need to for an additional device to generate MIDI. I often use a Peavey PC 1600X (MIDI fader box) for this purpose; it is the smallest device in my studio with pedal inputs. While I have seen plans for a footswitch that generates its own MIDI output, the slight gain in convenience does not justify moving to a more esoteric piece of equipment.  A step up in complexity is a control voltage pedal. 1.2 MIDI wind controllersA full history of MIDI wind controllers can be found at the International Wind Synthesis Association's site (see link below). Yamaha's WX series is the most common system and is still in production. All of the WX instruments are MIDI controllers based on a modified saxophone fingering system, and incorporating breath pressure and embouchure sensors. I was lucky to have acquired a WX-7 right after that model was discontinued, and this is the model I play now. Yamaha replaced it with the less able WX-11, which is also out of production. The current model, the WX-5, seems to be the best of the bunch. It is basically similar to the WX-7, and many of the augmentations described below should be relevant to the WX-5, too.  Whereas an interested performer can master a foot switch or a CV pedal within a few hours, picking up a MIDI wind controller should be thought of a learning a new instrument. Wind Controller Pedagogy will be a separate paper; this document will simply describe the control possibilities. Hopefully, that will be enough to encourage performers and composers to dig more deeply. The WX-7 has a number of physical sensors which are mapped to specific MIDI messages. There are two continuous controllers located in the mouthpiece. One measures wind pressure -- that is, how hard is the performer blowing. This measurement can be set to send out as breath control (continuous controller 2) or MIDI volume (cc 7), and also influences note-on velocity. Unless I'm programming a percussive sound, I usually ignore note-on velocities, and link amplitude and timbre directly to breath control data. This allows a full range of expression, and more accurately represents the response of a wind instrument. As an illustration, a crescendo from niente to fortissimo will be limited if the timbral quality of the patch is still referring to an extremely low note-on velocity. Since this sensor only senses pressure, the air in the instrument does not have to be moving forwards, or at all. Plugging up the exit at the bottom of the instrument makes circular breathing very simple. Or a balloon can be fit over the end, for a type of MIDI bag pipe -- perhaps more interesting as a theatrical effect than a musical modification. The embouchure sensor is a very interesting feature, although poorly implemented on the WX-7. The mouthpiece has a simulated reed, and the deflection angle is measured and sent out as pitch bend data. Unfortunately, there is a small pitch bend wheel under the right thumb that is also sending out pitch bend data. Individually, both controllers have a lot of possibility. But the fact that they send out the same data means that they are not as useful as they should be. (I understand this has been fixed on the WX-5.) Both are fairly difficult to control in a precise fashion, but are very functional for timbral control. Using the embouchure sensor for microtonal control is also effective, because it applies the performer's natural instincts. The keys are predictable, being another variant of the Boehm System. There are some useful trill fingerings and alternate fingerings, but ultimately they are a familiar means of sending out seven octaves of MIDI note numbers. The keys are just switches, so there is no innate possibility of half-hole techniques (such as on a clarinet or an open-holed flute) or aftertouch. Gary Scavone's "Holey" controller (listed below) offers some interesting ideas for augmentation in this area. A button beneath the right thumb allows a note to be sustained while other notes are played, or sets an interval to be played in parallel motion. This persists until cancelled. Finally, through a combination of octave keys and another button below the right thumb, program changes 0-4 can be sent. 1.3 trackingA great deal of performance information can be captured using a microphone. 2. augmentations2.1 design principals
2.2 sensorsIn order to access performance gestures, some basic technology is required to translate these gestures into a format that a computer or synthesizer can understand. Typically, the technology is represented by a pair of transducers. The first transducer translates a physical dimension -- force, position, acceleration, heat, etc. -- into a variable voltage. There are numerous sensors with possible performance applications; essentially any physical parameter that can be measured can be used. For instance, Laetita Sonami uses a combination of touch, flex and proximity (ultra-sound) sensors in her performance with the "Lady's Glove." Infusion System's web site (below) list a broad range of pre-fabricated sensors for use with their I-Cube system. The fields of robotics and security offer diverse solutions to the problem of moving data form the physical world to the computer. In keeping with my design principals, I choose to work with sensors that are robust, inexpensive, and relatively covert. This paper discusses the application of force sensing resistors, accelerometers, and position sensing ribbon controllers. 2.3 analog to digital convertersThe analog voltage from the chosen sensor needs to be converted to a digital signal. There are a number of commercially available analog-to-digital converters. Some systems specifically designed for artists and performers; these are relatively expensive but require less assembly and technical knowledge. Examples include Infusion Systems I-Cube and IRCAM's AtoMic. Alternately, a system can be built around an inexpensive microcontroller (e.g. BASIC Stamp). Many performers are using custom fabricated alternative controllers with great success. Such devices require some skill with electronics, or assistance of an institution such as STEIM. Numerous documents exist on the web that deal (see resources, below) with this technology, which are worth exploring. While many musicians may be intimidated by the prospect of soldering circuit boards, the advantages of this approach are manifold: for a lot less money, you can build exactly what you need. And, if you built it you can repair it. This paper focuses on commercial hardware that could be used by a musician with limited fabrication resources. The Gassmann Studio owns an I-Cube System that converts the signals of a diverse range of analog sensors into MIDI data that can be brought through a MIDI interface into Max/MSP. While the design principles in this paper are implemented using the I-Cube, they are applicable to any technological situation that suits the performer's skills and means. 2.4 flute augmentationThe Boehm System Flute is truly a wonderful instrument. In the hands of a skilled player it is amazingly facile, and has a very even, pure timbre across it's entire range. Unfortunately, this can be a liability if one is trying to explore a broad range of timbre. I have always been a little bit envious of brass's mutes, string's sul ponticello, electric guitars (!), and so on. This feeling, along with the flute's great agility, is one of the reasons that the instrument is at the forefront of the extended technique movement. (Examples are numerous, starting with Varese and Berio, and including work by Robert Dick and Kaija Saariaho.) It is clear that flute players and composers are searching for new sonic resources. Surprisingly, there is no MIDI control option available to flutists, sort of learning single reed technique for Yamaha's WX controllers. Therefore, the flute became my first candidate for augmentation. Matthew Burtner's Metasax employ's force sensing resistors (FSR's) on the keys of a tenor saxophone to give the performer a control similar to aftertouch on a MIDI keyboard. This strategy has a number of advantages. It relies on the fingers in their natural position on the instrument. There is an amount of haptic feedback -- the performer can feel the pressure. And, there is a certain "expressive logic" to controlling timbre by squeezing the instrument. Finally, these sensors are readily available in many sizes and are quite robust. Application of sensors to the keys of the flute presents some special challenges. First, I am working with an open-holed flute, which is necessary for many contemporary performance techniques. As the name suggests, the keys that are directly depressed (A, G, F, E, and D) are perforated, with an outer ring and an open center. For effects such as pitch bends and alternate timbres, a flutist might slide from completely covering the center hole, to partially covering, to closing the rim only. As such, the relationship of the finger to these keys is mutable, and the performer's tactile sense is very specific. It would be difficult to retrofit sensors to these keys that obtained reliable readings without interfering with the performer. For these reasons, I decided to focus keys activated by levers. 
Another potential problem appears when applying FSR's to a curved surface -- essentially any part of the flute. FSR's are constructed of the following components: resistive material applied to a film, a spacer adhesive, and another film printed with interdigitating contacts. The resistive material creates an electrical path between the contacts; pressure on the FSR creates a better contact, decreasing resistance. Application to a curved surface would preload the sensor, limiting its range. One goal of future research is to experiment with different sizes and placements of sensors, to determine how critical this effect is. 2.5 WX-7 augmentation***** 3. future research***** references
updated: 16 February 2003 | ||
With the smaller, square sort
I occasionally wonder if I successfully depressed it, leading to an
uncomfortable choice: press it again or not? (This is a case where visual
feedback -- seeing the computer screen -- is very helpful. A monitor near the
performer is often a workable situation, but can also be distracting for the
audience.) Also, the large pedals have more convenient surface area for taping
to the stage. The large body can be a drawback if the performer is unsure of
the location of the pedal. I have seen performers step on the body of the
pedal, and miss a cue. For this reason, some performers prefer the square
design -- the top part is all pedal. It is always wise to check with other
performers to see what they prefer.