Beitrag zur automatischen Erkennung gesprochener Ziffern

Some experiments concerning the utility of a simple function-model of the human ear for automatic speech recognition are reported. A short description of this function-model which was built by Zwicker, is given. Using a simple recording and evaluation method it is demonstrated, how the channel-time-patterns obtained from the function-model depend on the properties of different speakers. The German numbers "Null, Eins, Zwo, Drei, Vier, Fünf, Sechs, Sieb, Acht, Neun", spoken by 48 speakers are used as speech material.