October 26, 2016
|Time:||11:00 - 13:00|
Marco Matteo Markidis • Giuseppe Silvi
path to path~
This extended demo describes path~, a Pure Data external implementing a corpus-based, concatenative analysis-synthesis engine where sound is acquired and synthesized by a similarity match between audio grains, using a set of audio features extracted offline for sound corpora and in real-time for live input. The basic idea of this work is to segment the audio files, to perform a feature extraction analysis and to construct a k-d tree and a list of k-nearest neighbors for every grain, where every grain is represented as a set of points. In real-time, features are extracted from a live input and the nearest neighbor is found in the tree. Starting from this element and using its k-nn list, a set of grains is available for the synthesis. The main focus of this work is the real-time audio mosaicing. This presentation also describes how path~ is used for Giuseppe Silvi’s S4Ǝꟻ composition for Augmented Saxophone and Live Electronics. This work is written considering path~ the main algorithm for digital signal processing, sound analysis and synthesis in Pure Data.