Spectrum Analysis for Practitioners of Auditory Intervention Techniques
Wayne J. Kirby, B.M., M.M., D.A.
University of
North Carolina at Asheville
Introduction Virtually all forms of therapy involving auditory stimuli require an understanding of frequency spectrum analysis concepts and techniques. The purpose of this article is to present a brief overview of what spectrum analysis is, how it is done, and where to find some appropriate software.
Spectral content of music and sound According to the 19th Century French mathematician, Jean Baptist Fourier, all sounds comprised of periodic sound waves can be recreated by adding a series of sine waves together at varying amplitudes (what we perceive as "intensity" or "loudness"). Spectrum analysis is a method of identifying the frequencies (and their corresponding amplitudes) present in a sound by performing a computational process known as a Fast Fourier Transform.
The importance of spectral
analysis Practitioners of auditory intervention techniques should be
aware of the spectral content of the stimuli they are delivering to the auditory systems
of their clients, students, trainees or patients. This is particularly important for
professionals who practice the Berard and Kirby Methods of Auditory Integration Training
(AIT) as well as many other sound- and music-based interventions.
It has been theorized that broad spectrum sound is particularly important in creating AIT stimuli for several reasons. Modulating sound by dynamically modifying the relative amplitude of key spectral components of music appears to stimulate the entire auditory system, including the conductive, sensorineural and retrocochlear mechanisms. Additionally, insuring the presence of the appropriate spectral content allows for effective notch-filtering of masking frequencies. This filtering procedure facilitates the subject's perception of the otherwise masked (and therefore unperceived) stimuli. Masking phenomena may account for the auditory perception anomalies of some subjects. For example, a person who has a significant hypersensitivity to a particular frequency may be unable to perceive neighboring frequencies within the corresponding critical band due to amplitude masking. Without ascertaining, through spectral analysis, that the appropriate spectral content is contained in the music, the potential benefits gained from notch-filtering hyper-sensitive frequencies can not be realized.
Spectral Analysis techniques
Real Time Analyzers (RTAs): These devices are commonly used by
professionals in the audio industry. They are dedicated hardware-based digital audio
processors. The KAM PC-based auditory delivery system I developed for AIT
practitioners includes the capability of viewing spectral content in real time using this
kind of digital audio processor.
Non-Real Time Analyzers: These systems are
generally computer-based. In essence, they are data acquisition systems that can perform
very complex mathematical processes on the acquired audio data. Usually, the sound to be
analyzed is recorded into the computer by connecting the output of the CD player, AIT
device or other audio component directly to the input of the computer's sound card using
the appropriate audio adaptors. The signal is recorded onto the hard drive of the PC. The
stereo signal can then be viewed on the computer screen as a waveform graph representing the varying overall
intensity of the sound over a specific period of time (FIGURE 1).
The user may select a portion of the waveform they wish to analyze. The computer then performs the mathematically complex Fast Fourier Transform (FFT) operation on the selected waveform region. This process converts the visible waveform graph representing sound in the time domain to either a spectrum graph (FIGURE 2) or a sonogram (FIGURE 3). Both types of graphs represent the waveform graph's conversion into the frequency domain. The resulting graphs depict the intensity of each frequency over a specific time period.
Available software There are a number of computer programs available for spectrum analysis. I often use the spectrum analyzer function in Sound Forge, a professional audio software package manufactured by Sonic Foundry, Inc. However, there are a number of other programs available over the internet that are either shareware or freeware. You can find links to a number of these programs at http://www.hitsquad.com/smm/cat/SPECTRUM_ANALYZERS. This site provides download links to a wide variety of spectrum analysis programs for MacIntosh, Windows, Linux, Atari and other operating systems. One popular software program that contains a spectrum analysis function is Cool Edit. You can download an early shareware version of Cool Edit '96 at the link provided below. This program is designed to be run under Windows 9x. This software will allow you to record the music you wish to analyze on your hard drive via your sound card directly from your CD player, AIT device or other audio component. You can then analyze various sections of the sound sample by clicking on the desired points in time represented on the waveform graph. You can also set the software to show the fluctuating amplitudes of the spectral content in real time (as the sound sample is playing). If you find this demonstration program useful, you can obtain additional information on a fully functional version of the software from Syntrillium Software Corporation at http://www.syntrillium.com.
Conclusion Knowledge of spectrum analysis theory and practice
is essential to the effective application of the many varieties of auditory intervention
techniques currently being practiced. I hope that this article will help to convey the
value of objective measurement and analysis of the auditory stimuli that form the basis of
this continually-evolving music technology.
Figure 1 (below): Example of a
waveform graph. X axis = time; Y axis = amplitude (Time Domain)
Figure 2 (below): Example of a spectrum graph of the higher-amplitude, middle portion of the waveform shown above. X axis = frequency; Y axis = amplitude (Frequency Domain)
Figure 3 (below): Example of a sonogram of the the higher-amplitude, middle
portion of the
waveform shown in Figure 1. X axis = time; Y axis = frequency; amplitude represented
by darkness (Frequency Domain)
About the Author
Wayne J. Kirby holds the rank of full professor at the University of North
Carolina at Asheville where he specializes in psychoacoustics and music technology. Dr.
Kirby holds degrees from The Juilliard School, Yale University, and New York University.
He has published numerous articles on AIT, music technology and related topics. He can be contacted at
Serious Composer, Inc. PO Box 18041, Asheville, NC 28814. Telephone:
828-216-5629; Fax: 828-253-4573; email:
DrKirby@KirbyAit.com
Copyright 2000 Wayne J. Kirby.
All Rights Reserved Worldwide.
No part of this publication may be reproduced in any form or by any means without prior
written permission of the copyright owner. International copyright secured.