dbogdanov

Building Essentia's Python extension on Windows

Python.org Windows binaries are compiled against an MS Visual C++ (MSVC) runtime, which version differs with different Python versions. Essentia's Python extension should be built with MSVC too because the MinGW C++ library builds won't be compatible with MSVC.

In this guide we use VirtualBox Windows machine.

Preparing build environment

• Download Dev Virtual Machine: https://developer.microsoft.com/en-us/windows/downloads/virtual-machines
• Remove Visual Studio 2017 Community as we won't use it
• Install Build Tools for Visual Studio 2017: https://www.visualstudio.com/downloads/#build-tools-for-visual-studio-2017

dbogdanov

Task 1: correlation between descriptors and music categories

You are given a dataset of descriptors for music recordings and additional annotations by genres and moods.

• Analyze global correlations (across all music recordings in the dataset)
  • between descriptors within Group 1
  • between descriptors within Group 2
• Analyze a correlation of descriptors in Group 1 and Group 2 with
  • music genres associated with the recordings
• music moods (mood_happy, mood_sad, mood_agressive, mood_relaxed, mood_party)

dbogdanov

{
"lowlevel": {
    "average_loudness": 0.982675015926,
    "barkbands_crest": {
        "dmean": 3.16559839249,
        "dmean2": 5.15330886841,
        "dvar": 8.76487731934,
        "dvar2": 19.6714382172,
        "max": 24.5472221375,
        "mean": 11.2651510239,

dbogdanov

Task 1: Emotion regression

Description

Train and evaluate two Support Vector Regression models for prediction of arousal and valence. Use Essentia's MusicExtractor to compute summarized descriptor values (no frames). Use Scikit-learn for Support Vector Regression. It will be useful to pre-process all features first, standardizing them to zero mean and unit variance.

Write a report in form of a python notebook. This script should:

• Analyze all audio in the dataset using MusicExtractor and store aggredated analysis results in json files
• Load descriptors computed for all files, and load arousal/valence annotations
  • You can first convert json to CSV if you like
• Use lowlevel.*, rhythm.*, and tonal.* descriptors

dbogdanov

Dmitry Bogdanov

Post-doctoral researcher at Audio Signal Processing Lab at Music Technology Group, Departament de Tecnologies de la Informació i les Comunicacions, Universitat Pompeu Fabra.

• Address: Roc Boronat, 138, 08018 Barcelona
• Phone: (+34) 935422176
• [email protected]
• https://es.linkedin.com/in/dibogdanov

Research interests

Music information retrieval, sound and music computing, audio analysis, music recommendation, user modeling, machine learning, data mining, personalization.

dbogdanov

#include "algorithmfactory.h"
#include "algorithms/rhythm/rhythmextractor2013.h"
#include "algorithms/extractor/rhythmdescriptors.h"
#include "algorithms/tonal/vibrato.h"
#include "algorithms/standard/stereodemuxer.h"
#include "algorithms/tonal/pitchcontoursmonomelody.h"
#include "algorithms/io/audiowriter.h"
#include "algorithms/sfx/maxtototal.h"
#include "algorithms/tonal/harmonicpeaks.h"
#include "algorithms/spectral/maxmagfreq.h"

dbogdanov

import numpy as np
import essentia.standard
import matplotlib.pyplot as plt

 
def sinewave(f0, nseconds, samplerate=44100):
    t = np.arange(1, samplerate * nseconds)
    sig = np.zeros_like(t)
    return sig + 0.5 * np.sin(2 * np.pi * f0 * t / samplerate)