Browse Prior Art Database

Pattern Analysis for Musical Melodies

IP.com Disclosure Number: IPCOM000040606D
Original Publication Date: 1987-Dec-01
Included in the Prior Art Database: 2005-Feb-02
Document File: 2 page(s) / 14K

Publishing Venue

IBM

Related People

Pickover, CA: AUTHOR

Abstract

A technique is described whereby musical melody patterns are analyzed utilizing spectrograms and three-dimensional power spectrum represent- ations, similar to those used in speech analysis. The analysis focuses on the mathematical characterization of melody patterns using moving data windows and interactive graphic computer programming systems with a wide variety of controlling parameters. Sound is not used as input. The concept is not confined to melodic sequence analysis, but can be expanded utilizing characterized periodicities in a data set.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

Pattern Analysis for Musical Melodies

A technique is described whereby musical melody patterns are analyzed utilizing spectrograms and three-dimensional power spectrum represent- ations, similar to those used in speech analysis. The analysis focuses on the mathematical characterization of melody patterns using moving data windows and interactive graphic computer programming systems with a wide variety of controlling parameters. Sound is not used as input. The concept is not confined to melodic sequence analysis, but can be expanded utilizing characterized periodicities in a data set. For example, the concept may be expanded to include studies for the treat- ment of a human bladder cancer gene as a melodic sequence in order to detect periodicities, the analysis and synthesis of computer singing the combined voices of a chorus and the representation of periodicities in the movement of a breathing protein. The two methods of describing patterns of melodic sequence are the three-dimensional power spectrum and the spectrogram, which focus on the comparisons of intensity, frequency and position of the sequence of notes. The spectrogram accepts as input a coded version of the musical score and subsequently computes digital spectrograms and topographic spectral distribution functions.

The power spectral analysis of instrument sounds utilize algorithms which are sensitive to the periodicities in the melodic pitch sequence and do not have as an input a traditional acoustic sound source. The computer does not analyze time wave forms, therefore parameters, such as loudness, attack and timbre are not characterized. Emphasis is placed on analysis of the frequencies of the progression of hills and valleys of the musical score, which of themselves represent fundamental frequencies of notes on a keyboard. Using a vector graphics display system, the topographic power spectra mathematically and visually portray the patterns of a melody on a three-dimensional map. The pattern analysis accepts as input the pitch values of the melody fluctuating in time. Musical notes are entered into text files using a convenient three-letter code which specifies the note, octave and whether the note is a sharp or a flat.

The program converts the alphanumeric input to pitch values using a loo...