Music Mood Classification using Python

Fullført Lagt ut 4 år siden Betales ved levering
Fullført Betales ved levering

Automatic recognition of emotions in musical audio has gained increasing

attention in the field of music information retrieval (MIR) during the past few

years. The development in the field has coincided with the need for managing

large collections of digital audio for the public via web services such as Spotify

and Last.fm. This is reflected, for example, in the increasing number of

submitted systems in the annual Audio Music Mood Classification (AMC)

contest part of the Music Information Retrieval Evaluation eXchange3

(MIREX). The substantial variance in the submitted systems in the contest

and in the approaches in the MIR field in general indicates a lack of consensus

concerning the choice of an underlying psychological model for emotions or

moods in music and the establishment of precise machine learning conventions.

Despite research in musicology studying the influence of specific cues in the

musical structure on emotional expressions, there is no complete analytical

consensus about the required “ingredients”—i.e., acoustical features extracted

from music—to build the optimal models for emotion recognition, in the limits

imposed by the subjectivity of [login to view URL] descriptors for each music track

provided for this activity are the state-of-the-art descriptors used in this area

(see table on next page for details). Your task is to determine which sets of

features are most useful for learning the accurate classifiers. In particular your

task isto explore different feature sets and machine learning algorithms with the

aim to obtain the most accurate classifier. As part of your exploration you are

supposed to get the corresponding accuracies for several algorithms (e.g.

SVM, KNN, Decision Trees, Neural Networks, Ensemble, Naive Bayes), and

applying training set, bootstrap, and cross-validation methods to get the

training and testing accuracies. What kind of features (timbre, rhythm, pitch,

dynamics, structure or harmony) are more relevant for predicting emotion in

music in the provided data set?

Python Programming

Prosjekt-ID: #22309917

Om prosjektet

4 bud Eksternt prosjekt Aktiv 4 år siden

Tildelt til:

mza5ab8dc9be9a8c

Note: I read all of your requirements and i can deliver you with quality work. I have been worked and implemented all the techniques which you have mentioned to be done. Greetings! Welcome to DEEP LEARNING, MACHINE LE Mer

€15 EUR på 1 dag
(64 omtaler)
5.6

4 frilansere byr i gjennomsnitt €15 for denne jobben

vaibhavAP

5 years experience in Reactjs / Redux / Angular / Nodejs / PHP / Django / Backned - Frontend development! All of our programming skills: *Front-end: - HTML, HTML5, JSON. - JAVASCRIPT (Ajax, AngularJS / 2 / 4 / 5 / Mer

€13 EUR på 6 dager
(0 Omtaler)
0.6