CCRMA Summer Workshop
Digital Signal Processing for Audio: Spectral and Physical Models (at Stanford University)
Perry Cook, Xavier Serra
This course covers analysis and synthesis
of sounds based on spectral and physical models. Models and methods for synthesizing real-world sounds as well
as musical sounds are presented. The course is organized into morning lectures covering theoretical
aspects of the models, and afternoon labs. The morning lectures present topics such as Fourier theory,
spectrum analysis, the phase vocoder, digital waveguides, digital filter theory, pitch detection, linear
predictive coding (LPC), high-level feature extraction, and various other aspects of signal processing of
interest in sound applications.
The afternoon labs are hands-on sessions using an implementation of SMS in Matlab, a recently developed set of libraries in C++ called CLAM, the Synthesis ToolKit (STK) in C++, and other software
systems and utilities. Familiarity with engineering, mathematics, physics, and programming is a plus, but the
lectures and labs are geared to a musical audience with basic experience in math and science. Most of the
programs used in the workshop are available to take home.
Given the short duration of the workshop and the broad spectrum of topics to cover, the lectures are necessarily
fairly high level in nature. However, a full complement of in-depth readings will be provided for those who
wish to investigate the details of the material. Also, the last two days of the workshop include a more
detailed treatment of some advanced topics and the corresponding afternoon labs give the students a chance
to solve some specific problems of their interest.