Computational Modelling of Paralinguistics: Overview
Björn Wolfgang Schuller, Anton M. Batliner · 2013
This chapter provides an overview on the usual steps and building blocks involved in computational modelling of paralinguistics. The chapter describes the steps such as preprocessing, low-level descriptor extraction, and optional steps such as chunking, and (Hierarchical) functional extraction. It outlines the feature reduction, and the training/adaptation phase of feature selection/generation, parameter selection, and model learning. The chapter gives a detailed account of the classification/regression, and optional steps such as fusion, encoding, and the training/adaptation phase of audio databases. Finally it describes the acoustic model (AM) and language model (LM).