Automated thresholding in electric guitar note recognition using Fourier analyis and neural network

• Main Authors: Tensuan, J.V. S., Villagracia, A., Pobre, Romeric F.
• Format: text
• Published: Animo Repository 2005
• Subjects: Fourier analysis; Sound; Physical Sciences and Mathematics; Physics
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/12382
• Institution: De La Salle University

Writing manually the musical notes on a tablature while playing a musical instrument is time- consuming for the musicians in composing songs. To resolve this problem, this research paper generates a real-time automated transcription system that will write the musical pieces on a computer- generated tablature while playing the guitar. This is done through recording the wave data from an electric guitar by a computer and undergoes fourier transform. The transformed data is fed into a neural network which is trained to recognize the pattern of the data using the backpropagation algorithm. Each network is assigned to different chords to recognize them respectively. Given the input data, the network will automatically compute for a threshold value which can range from 0.8 to 0.9 to determine the musical note from the trained neural network. Finally, the automated transcription software which was compiled and built in Microsoft Visual Basic and C++ will display the played notes in a tablature as the musician plays the guitar.