Assignments for Audio Signal Processing for Music Applications on MTG-UPF Msc Sound And Music Computing.
IMPORTANT NOTE: This repo is intended for my personal learning purpose only.
This exercise aims to get familiar with some basic audio operations using Python.
Doing this exercise you will get a better understanding of the basic elements and operations that take place in the Discrete Fourier Transform (DFT).
With this exercise you will get a better understanding of some of the Fourier theorems and of some useful properties of the DFT.
Doing this exercise you will learn about the concept of the main lobe width of the spectrum of a window and you will better understand the short-time Fourier transform (STFT). You will also use the STFT to extract basic rhythm related information from an audio signal, implementing an onset detection function, which is one of the rhythm descriptors often used in music information retrieval to detect onsets of acoustic events.
In this exercise you will experiment with the sinusoidal model, measuring and tracking sinusoids in different kinds of audio signals. You will use the sinusoidal model to analyze short synthetic sounds with the goal to better understand various aspects of sinusoid estimation and tracking. You will experiment with different parameters and enhancements of the sinusoidal modeling approach.
This exercise on the Harmonic model will help you better understand the issue of fundamental frequency estimation by analyzing several sound examples with harmonic content.
In this exercise you will analyze and synthesize sounds using the Harmonic plus Stochastic (HPS) model, hpsModel.py.
In this exercise you will use the HPS model to creatively transform sounds. There are two parts in this exercise. In the first one you should perform a natural sounding transformation on the speech sound that you used in the previous exercise (E7). In the second part you should select a sound of your choice and do a "creative" transformation. You will have to write a short description of the sound and of the transformation you did, giving the link to the original sound and uploading several transformed sounds.
##Week 9: Sound and music description With this exercise you will learn to describe sounds with simple machine learning methods. You will learn to use the Freesound API to load pre-computed sound descriptors from Freesound and to perform sound clustering and classification with them. You will work with instrumental sounds, thus learning what audio features are useful for characterizing them.