Digitalized Sound Spectrograph Using FFT and Multiprint Techniques
E. H. Rothauser, Dietrich Maiwald · The Journal of the Acoustical Society of America · 1969
In the history of speech research, sound spectrographs are playing the rôle of an almost invaluable tool. Besides the many advantages of these instruments, some of their limitations have also become evident. Quite a few of these are a direct consequence of the utilization of analog techniques in these devices. The present paper tries to show that digital-processing techniques allow to implement a spectrographic system much more powerful than its analog counterpart. Its hardware consists of a general-purpose computer adapted to accept digitally encoded speech signals, and a standard chain printer. The short-time power spectra are computed utilizing the fast Fourier transform algorithm. Spectrograms with eight gray levels can be generated by multiprinting adequate sequences of characters out of the basic 48-character set. The necessary computer programs are sufficiently simple to be easily tailored to the specific needs of any particular investigation. Wherever digital processing is used in speech research, the approach described exhibits its greatest advantages. The digital spectrograph can then be seen to constitute an organic part of the total software system being a subroutine that may be called whenever needed.