Computational Bioacoustics

6 Two-dimensional audio parameterization

Introduction

A new paradigm for audio pre-processing and parameterization, which makes use of image processing techniques, was recently introduced. A common feature of all methods grounded on this paradigm is that these compute the audio spectrogram, which is then considered a greyscale image, and image processing techniques are applied in order to implement noise suppression, audio parameterization, or both. These methods always operate simultaneously on multiple columns and rows of the spectrogram, that is, take advantage of the concurrence of temporal and frequency information and by that reason are referred to as 2-D audio parameterization methods. The simultaneous operation along the temporal and ...