Impact of Vocal Tract Length Normalization on the Speech Recognition Performance of an English Vowel Phoneme Recognizer for the Recognition of Children Voices

Differences in human vocal tract lengths can cause inter speaker acoustic variability in speech signals spoken by different speakers for the same textual version and due to these variations, the robustness of a speaker independent (SI) speech recognition system is affected. Speaker normalization using vocal tract length normalization (VTLN) is an effective approach to reduce the affect of these types of variability from speech signals. In this paper impact of VTLN approach has been investigated on the speech recognition performance of an English vowel phoneme recognizer with both noise free and noisy speech signals spoken by children. Pattern recognition approach based on Hidden Markov Model (HMM) has been used to develop the English vowel phoneme recognizer. Here training phase of the automatic speech recognition (ASR) system has been performed with speech signals spoken by adult male and female speakers and testing phase is performed by the children speech signals. In this investigation, it has been observed that use of VTLN can effectively improve the robustness of the English vowel phoneme recognizer in both noise free and noisy conditions.