Time and Pitch Scale Modification of Audio Signals

Abstract

The independent control over the time evolution and the pitch contour of audio signals has always stood very high on the wish-list of the audio community because it breaks the traditional tie between pitch and rate of playback. Rudimentary solutions to that problem date back to the analog days, but the recent advent of fast digital audio processors has made it possible for better methods to make their way into consumer products, with applications in karaoke, phone answering-systems, post-production audio/video synchronization, language learning, to name only a few. This chapter describes several popular methods for the time-scale, pitch-scale and formant-scale modification of audio signals, and summarizes their respective strengths and weaknesses as well as their relative costs in terms of computation power.