Interactive Computer Music Systems

A session of the Acoustical Society of America (ASA) conference in Pittsburgh, PA,
Tuesday Morning, 8:25-11:30am, June 4, 2002,
chaired by Roger B. Dannenberg, School of Computer Science, Carnegie Mellon University, Pittsburgh

This session explores computer music performance systems and virtual instruments, especially the use of sensors and computing to control real-time sound generation processes. The presenters are an internationally distinguished collection of scientists, composers, and musicians, and most are confortable in two or more of these roles.

8:25 Chair's comments: Roger Dannenberg
8:30 AM

The structural implications of interactive creativity

Joel Chadabe, Electronic Music Foundation, Ltd.

The functioning of any particular electronic musical instrument can be placed somewhere along a line that extends from deterministic to indeterministic. Bearing in mind this author's conviction that the goal of technology should be to better human existence, we ask: In what ways does an electronic musical instrument function for the benefit of its performer? Although deterministic instruments may offer more powerful controls than traditional instruments, they typically put a performer in the traditional situation of making a gesture and expecting a predictable effect. Indeterministic instruments, on the other hand, put a performer in an interactive role of improvising relative to an unpredictable output. By 'interactive', I mean 'mutually influential'. The performer influences the instrument and the instrument influences the performer. The unique advantage of such interactive instruments is that they foster 'interactive creativity'.

The design of a traditional instrument is fundamentally different from the design of an interactive  instrument. A traditional instrument is structured as a single cause and effect, articulated as a synchronous linear path through a hierarchy of controls from a performer operating an input device to the multiple variables of a sound generator. An interactive instrument, on the other hand, is structured as a network of many causes and effects at various levels of importance, with a performer's input as only one of the causes of the instrument's output in sound.

The author will present several historical examples of interactive electronic musical instruments and offer some speculations on the future.


9:00 AM

Music scene description: Toward audio-based real-time music understanding

Masataka Goto, PRESTO, JST. / National Institute of Advanced Industrial Science and Technology (AIST, Japan)

Music understanding is an important component of audio-based interactive music systems. A real-time music scene description system for the computational modeling of music understanding is proposed. This research is based on the assumption that a listener understands music without deriving musical scores or even fully segregating signals. In keeping with this assumption, our music scene description system produces intuitive descriptions of music, such as the beat structure and the melody and bass lines. Two real-time subsystems have been developed, a beat tracking subsystem and a melody-and-bass detection subsystem, which can deal with real-world monaural audio signals sampled from popular-music CDs. The beat tracking subsystem recognizes a hierarchical beat structure comprising the quarter-note, half-note, and measure levels by using three kinds of musical knowledge: of onset times, of chord changes, and of drum patterns. The melody-and-bass detection subsystem estimates the F0 (fundamental frequency) of melody and bass lines by using a predominant-F0 estimation method called PreFEst, which does not rely on the F0's unreliable frequency component and obtains the most predominant F0 supported by harmonics within an intentionally limited frequency range. Several applications of music understanding are described, including a beat-driven, real-time computer graphics and lighting controller.

9:30 AM

Making the computer ``listen'' to music

Christopher Raphael,   Dept. of Mathematics and Statistics,  University of Massachusetts, Amherst

A computer system is discussed that provides real-time accompaniment to a live musician playing a non-improvisatory piece of music.  Particular attention is devoted to the ``listening'' process, in which the computer must follow the soloist's progress through the musical score by interpreting the sampled acoustic signal.  The process is complicated by the significant variation and occasional errors from the live player during performance. A hidden Markov model is introduced providing a principled, trainable and fast solution to the listening problem.  The system is capable of assessing its own level of uncertainty about score position, as well as accomodating the sometimes strong signal component from the accompaniment instrument.  A live demonstration will be provided.

10:00 AM

Machine musicianship

 Robert Rowe,  Department of Music and Performing Arts Professions, School of Education, New York University

The training of musicians begins by teaching basic musical concepts, a collection of knowledge commonly known as musicianship. Computer programs designed to implement musical skills (e.g., to make sense of what they hear, perform music expressively, or compose convincing pieces) can similarly benefit from access to a fundamental level of musicianship. Recent research in music cognition, artificial intelligence, and music theory has produced a repertoire of techniques that can make the behavior of computer programs more musical. Many of these were presented in a recently published book/CD-ROM entitled “Machine Musicianship”. For use in interactive music systems, we are interested in those which are fast enough to run in real time and that need only make reference to the material as it appears in sequence. This talk will review several applications that are able to identify the tonal center of musical material during performance. Beyond this specific task, the design of real-time algorithmic listening through the concurrent operation of several connected analyzers is examined. The presentation includes discussion of a library of C++ objects that can be combined to perform interactive listening and a demonstration of their capability.

10:30 AM

The IMUTUS interactive music tuition system

George D. Tambouratzis, Stelios Bakamidis, Ioannis Dologlou, George Carayannis, Markos Dendrinos, Institute for Language and Speech Processing, Greece

This presentation focuses on the IMUTUS project, which concerns the creation of an innovative method for training users on traditional musical instruments with no MIDI (Musical Instrument Digital Interface) output. The entities collaborating in IMUTUS are ILSP (co-ordinator), EXODUS, SYSTEMA, DSI, SMF, GRAME and KTH.

The IMUTUS effectiveness is enhanced via an advanced user interface incorporating multimedia techniques. Internet plays a pivotal role during training, the student receiving guidance over the net from a specially-created teacher group. Interactiveness is emphasised via automatic-scoring tools, which provide fast yet accurate feedback to the user, while virtual reality methods assist the student in perfecting his technique. IMUTUS incorporates specialised recognition technology for the transformation of acoustic signals and music scores to MIDI format and incorporation in the training process.  This process is enhanced by periodically enriching the score database, while customisation to each user's requirements is supported.

This work is partially supported by European Community under the Information Society Technology (IST) RTD programme. The authors are solely responsible for the content of this communication. It does not represent the opinion of the European Community, and the European Community is not responsible for any use that might be made of data appearing therein.


11:00 AM

Interactive systems research at CNMAT

David Wessel, CNMAT, UC Berkeley

A live-performance musical instrument can be assembled around current lap-top computer technology. One adds a controller such as a keyboard or other gestural input device, a sound diffusion system, some form of connectivity processor(s)   providing for audio I/O and gestural controller input, and reactive real-time native signal processing software.   A system consisting of a hand gesture controller; software for gesture analysis and mapping, machine listening, composition, and sound synthesis; and a controllable radiation pattern loudspeaker are described.  Interactivity begins in the set up wherein the speaker-room combination is tuned with an LMS procedure.  This system was designed for improvisation.  It is argued that software suitable for carrying out an improvised musical dialog with another performer poses special challenges.  The processes underlying the generation of musical material must be very adaptable, capable of rapid changes in musical direction.   Machine listening techniques are used to help the performer adapt to new contexts.  Machine learning can play an important role in the development of such systems.  In the end, as with any musical instrument, human skill is essential.   Practice is required not only for the development of musically appropriate human motor programs but for the adaptation of the computer-based instrument as well.