CHORD ESTIMATION USING CHORD TEMPLATES AND HMM

This document describes a submission to the Music Information Retrieval Evaluation eXchange in the Audio Chord Recognition task. A set of chroma vectors representing the pitch content of the audio file over time is extracted. From these observations the chord progression is then estimated using hidden Markov models. The system takes into account the presence of higher harmonics of pitch notes and includes some music knowledge. In this approach, no training is needed.