"In 1930 a Hungarian scientist, Tihamer Nemes, filed a patent application in Germany for the principle of making an optoelectrical system automatically transcribe speech. His idea was to use the optical sound track on a movie film as a grating to produce diffraction patterns (corresponding to speech spectra), which then could be identified and typed out. The application was turned down as “unrealistic.” Since then the problem of automatic speech recognition has occupied the minds of scientists and engineers, both amateur and professional."
(Teuvo Kohonen, IEEE March 1988)
No comments:
Post a Comment