ReCHORD: a hardware interface for algorithm-based chord recognition

• Main Authors: Dimaranan, Baron A., Luy, Wilbert G., Tan Seng, Gero Christian C., Teves, Jude Michael M.
• Format: text
• Language: English
• Published: Animo Repository 2016
• Subjects: Chords (Music) > Data processing; Guitar > Chord diagrams > Data processing; Programming Languages and Compilers
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/6477
• Institution: De La Salle University

Chord recognition today continues to suffer from low accuracy with complex chords, with the majority of implementations relying, mostly on software alone. The music information retrieval evaluation exchange (MIREX), which has entries focused on software-based chord recognition. In MIREX 2015 for example, the latest algorithms presented for chord recognition resulted in 43.82% to 82.20% for basic chords, and 17.04% to 76.04% for complex chords. On the other hand, an approach using stereo cameras achieved an accuracy of 98.4%, though the system outputs a chord even if the sound produced is incorrect or none is produced at all, as long as the finger position is correct. In this research, a real-time chord detection system was built to recognize basic and complex chords from a guitar. With the use of a custom-built individualized string pickup, processing of a very complex signal due to the summation of all the signals from the 6 strings of the guitar would not be needed anymore. The software tailored for the hardware achieved 100% accuracy for note recognition and single-play chord recognition, and 91% continous-play chord recognition with the tempo slower than the temporal resolution of the system. The range of chords that were tested include major, minor, diminished, augmented, major 7th, minor 7th, diminished 7th, dominant 7th, suspended 2nd, and suspended 4th. With the Arduino as the microcontroller interface and all the software delay, a temporal resolution of around 2.2 seconds for the system is achieved.