Formant Tracking by Computer
C. Gordon Bell, Fausto Poza, Κ. Ν. Stevens · The Journal of the Acoustical Society of America · 1959
At the last meeting of the Acoustical Society of America a method of formant tracking using a digital computer was described. By this method synthetic speech spectra, constructed within the computer from a catalog of elemental spectra, are compared with measured spectra of a speech signal. The present paper discusses the evaluation, extension, and refinement of this method of analysis. Computer techniques which result in the display of a “synthetic sonagram” are described, together with examples of the use of the computer oscilloscope for displaying intermediate steps in the spectral matching process. Results obtained with a variety of speakers are presented. The problems of preparing a representative catalog of glottal spectra, evaluating the role of analyzing filter banks, and determining a meaningful “fit criterion” are considered. A method similar to the foregoing for synthesizing fricative spectra is also described. [[This work was supported in part by the U. S. Army (Signal Corps), the U. S. Air Force (Office of Scientific Research, Air Research and Development Command), and the U. S. Navy (Office of Naval Research); in part by Air Force Cambridge Research Center Contract AF 19(604)-2061; and in part by the National Science Foundation.]