Section: New Results

Embedding Audio Processing

Participants : Jean-Louis Giavitto, Pierre Donat-Bouillud.

Audio processing has been integrated in the Antescofo language. This experimental extension aims at providing sample-accurate control and dynamic audio graphs directly in Antescofo. Currently, FAUST (through a native embedding of the in-core compiler) and a few specific signal processors (notably FFT) can be defined. The tight integration enable specification of multiple-timed signal processing in conjunction with control programs. One example of this integration is the use of symbolic curve specification to specify variations of control parameters at sample rate, a task whose correctness in real-time is not at the scope of competing systems. Our approach has proven to provide such mechanisms at a lower computational cost; for example a factor of two in the remaking of Boulez' piece Antheme 2 compared to the original version with the audio effects managed in Max. We will further pursue such optimizations while extending sample accuracy, by developing a type-system to preserve block computations in case of preemptive audio processing  [41].

The reduced footprint enable the embedding of an Antescofo engine with internal audio processing on Raspberry PI and UDOO nano-computers (early results are reported in [26]).