Automatic chord estimation from audio through instrumental classification
Daniel Gil Ortiz
Degree on sound and image engineering
Polytechnic School. University of Alicante.
Extracting music information, specifically the automatic chord detection, is an open research area for many years because of the difficulty of performing this task manually. Thus, the development and implementation of an automatic system estimation chord from audio signals by comparing harmonic patterns (cromagrama) is proposed. The main objective for this work is to use specific harmonic patterns obtained by instrumental classification. For that, instrumental families have been studied, based of the General MIDI standard, and a clustering algorithm has been applied in order to look for better timbral patterns.
Technical report (in spanish)
Source code and data
A number of degree students are doing their academic works in our group
in diferent tasks related to pattern recognition and artificial intelligence research and development. You can
to know more about some of those projects.