Automatic chord estimation for using in musical audio transcription

Automatic chord estimation for using in musical audio transcription

Miguel Segura Sogorb

September, 2015

Degree in Image and Sound in Telecommunications

Escuela Politécnica Superior
Universidad de Alicante


The aim of this project is to implement a basic system of chord detection in music audio signals. From them, the chromagram is extracted and compared with a series of patterns that are characteristic of the chords to be estimated. The comparison is made through a similarity measure and different ways of creating the patterns from the harmonic content of the chords. Labelled databases have been used in order to assess the performance of the system, when comparing to three and four-note chords, in different ranges of pithches and dynamics, with dispare results.


Options

  1. Technical report (in Spanish)

[ Versión en castellano ]