Implementation of a Pitch Enhancement Technique: Punjabi Automatic Speech Recognition (PASR)

Automatic speech recognition (ASR) has been an active area of research for decades. As demand for ASR systems has been increased in the last few years due to their utility in a wide range of applications. The proposed work intends to develop a Punjabi ASR-based speech classification system using pitch enhancement corps optimization technique and Mel frequency cepstral coefficient (MFCC) feature extraction method. The complete experiment has been conducted on a total of 80 speakers containing comprising 6 hours of speech voice recording corpus. The classification language model resulted in the minimum word error rate (WER) of 8.2% in the case of the tri-gram model with 80 speakers. In addition to this, a comparison of the resulting outcomes has been conducted which present the outperformance of the tri-gram language classification model in the case of all set of speakers.