Design of digital filter bank by Kaiser windows to identify notes and guitar chords via binarized images with Otsu threshold 


In this paper is proposed a new procedure to identify notes and chords on guitar records. The approach mixes techniques of signal, audio and image processing and it is divided into five stages, namely: decomposition into energy components using a digital filter bank by Kaiser windows, conversion of image-to-matrix followed by an Otsu binarization threshold, adjusting from connectivity between pixels, composition and temporal overlap of the results. Tests run using guitar records (via synthesizer) were performed, obtaining accuracy rates exceeding 99%.