FORMANT ESTIMATION FOR SPEECH RECOGNITION

This paper presents a new method for estimating formant frequencies, T he formant model is based on a digital resonator. Each resonator repre sents a segment of the short-time power spectrum, The complete spectru m is modeled by a set of digital resonators connected in parallel, An algorithm based on dynamic programming produces both the model paramet ers and the segment boundaries that optimally match the spectrum. We u sed this method in experimental tests that were carried out on the TI digit string data base, The main results of the experimental tests are : 1) the presented approach produces reliable estimates of formant fre quencies across a wide range of sounds and speakers; and 2) the estima ted formant frequencies were used in a number of variants for recognit ion, The best set-up resulted in a string error rate of 4.2% on the ad ult corpus of the TI digit string data base.