| Popis: |
In this paper we present applications integrating two classic machine learning methods into a Computer-aided Composition environment with the specific purpose of notating, organizing and synthesizing audio from large sets of sound data. We present a modular sample replacement engine driven by a classification method, and a texture synthesis application employing a clustering method. The applications are designed and presented with a particular focus on modularity and extensibility, with the goal of providing flexible options for integration into existing OpenMusic projects. Therefore, in addition to presenting the metholodgy behind our applications, we also highlight the modular aspects of their structure along with several functions for performing transient detection, Mel-frequency cepstrum analysis, and probability vector calculation. |
