- William Brent -
Making Sense and Use of Audio Features in Creative Contexts
WHEN: Wednesday, December 4th, 6:30PM
WHERE: NYU Music Department
24 Waverly Place, Room 220, NY 10003
Here are the patches from the talk.
Certain standard audio features can be understood intuitively, but the meanings of others are less obvious. William Brent will discuss some practical and theoretical aspects of audio feature extraction in connection with his timbreID library for Pure Data. Special attention will be given to the meaning of more abstract feature information, such as cepstrum and Mel-frequency Cepstral Coefficients (MFCCs). Recent projects making use of such features will be reviewed as well, including a realization of John Cage’s Variations II that requires ordering of hundreds of audio files by timbre, and software for Bryan Christian’s Ignota, designed to compare the timbres of percussion instruments and sung IPA phonemes for the purpose of generating instrument-dependent language.
William Brent is a computer musician and Assistant Professor of Audio Technology at American University in Washington DC. His creative work is spread across the areas of experimental music performance, sound installation, and instrument design, and involves various combinations of human- robotic- and computer-realized sound. In collaboration with internationally recognized composers and performers, he develops and operates real-time audiovisual manipulation software for inter-media performance works. In this capacity, he has presented work at venues such as SESC (São Paulo), Glasgow Concert Halls (Scotland), Miller Theatre (New York), and the National Gallery of Art (Washington, DC). As a programmer, he develops open source software libraries for the Pure Data (Pd) programming environment. His current lines of research include new methods for physical control of synthesized audio, signal analysis techniques for quantifying timbre, and various aspects of human timbre perception.
Brent holds a Ph.D in Music from the University of California, San Diego, where he studied in the computer music area with Miller Puckette, F.R. Moore, and Shlomo Dubnov. Centered on various understandings of timbre, his dissertation research examined signal processing techniques for automatic classification of percussion instruments, and the relationship between objective measurements and human judgments of percussive sounds.
Software and further information can be accessed at www.williambrent.com.