Robust speech recognition with feature extraction using combined method of RSF and DRA

This paper explores the extraction of speech features aiming noise robustness for speech recognition and proposes advanced speech analysis techniques named RSF (Running Spectrum Filtering)/DRA (Dynamic Range Adjustment). The proposed techniques DRA and RSF focus on the speech feature adjustment. DRA normalizes cepstral dynamic ranges and RSF eliminates the jitter in. uences of speech feature parameters. The experiments on isolated word recognition were carried out using 40 male and female speakers each for training and 5 male and female speakers each for recognition. The results of recognition rate improving from 17% to 63% versus running car noise at -10dB SNR showed the effectiveness and high noise robustness of the proposed method.