Digital Filters, Filter Banks and Their Design for Audio Applications- With Python Examples

Presenter:
Gerald Schuller, Ilmenau University of Technology - IImenau, Germany; Fraunhofer Institute for Digital Media technology (IDMT) - Ilmenau, Germany

This tutorial will teach how to design "Finite Impulse Response" and "Infinite Impulse Response" filters for audio applications, in theory and practice, and will give examples in the popular Open Source programming language Python. Then it will go on to show how to design and use filter banks. Examples will be the "Modified Discrete Cosine Transform" (MDCT) filter bank, the (Integer-to-Integer) "IntMDCT," and Low Delay filter banks, which are widely used in MPEG audio coding standards. Further it will show digital filters as predictors for predictive coding, with applications in MPEG Lossless coding standards. Finally it will show how to implement filter banks as convolutional neural networks, which makes them "trainable" and eases the use of GPU's. This has applications for instance in audio source separation.