MATLAB Programming exercises of the Speech and Audio Signal Processing course

By Soroosh Tayebi Arasteh | سروش طیبی آراسته

This repository contains the programming exercises of the Speech and Audio Signal Processing course (SS 2020, Prof. Dr.-Ing. Walter Kellermann) offered by the Chair of Multimedia Communications and Signal Processing (LMS) of the Electrical Engineering Department at University of Erlangen-Nuremberg (FAU).

Context

• The main goal of the project is to illustrate different algorithms and to try them out in some real-world applications.

• The exercises are done in MATLAB.

• Every exercise contains a .pdf script file which explains the exercise's protocol.

• All the source files are provided by the Chair of Multimedia Communications and Signal Processing (LMS).

Overview of the project:

• Representation of Signals-General: Introduces the simulation language within a basic discrete-time signal processing context and goes on to the illustration of some important fundamentals on the representation of statistical signals. Linear mean, variance, white noise, normal distribution, ACF, PSD, basic filtering, pdf... . intermediate.m file contains the solutions for all the required exerimentations.

• Representation of Speech and Audio Signals: Analysis of speech and audio signals and basic estimation methods. Stationarity, Ergodicity, Long-Term and Short-Time Analysis of ACF and PSD, Periodogram, Welch and Bartlett methods, Pitch estimation.

• Speech Coding: Linear prediction: forward vs. backward, Psychoacoustics: the foundation for audio coding (e.g. MPEG audio).

• ASR 1-Fundamentals and Feature Extraction: Cepstral Analysis and Liftering

• ASR 2-Hidden Markov Models: Decoding feature vectors in order to obtain the text that was originally spoken. Isolated Word Recognition, Continuous Speech Recognition, Acoustic Modeling, Training of HMMs, Viterbi algorithm.

• Acoustic Human-Machine Interface: Acoustic Source Localization, Beamforming, Acoustic Echo Cancellation

In progress ....