Robust harmonic complex estimation in noise.
J.D Krijnders, Maria E. Niessen, T.C. Andringa · University of Groningen research database (University of Groningen / Centre for Information Technology) · 2007
Introduction Automatic speech recognition(ASR) systems work only in limited application ranges, because of noise, reverberations and speech variability. This poster addresses the noise problem. Two solution strategies exist: •Train the recognizer to ignore the noise • Separate the signal from the noise Here we focus on separating the signal from the noise. We do this based on fundamental frequency of the voiced parts of speech.