Feature extraction in enhancing speech signal for mobile communication

Mobile phones are considered as smart phones if it has additional characteristics other than regular wireless voice transmission. Smartness does not mean compromising with quality of voice under noisy conditions. This paper discusses about the methodology for improving the quality of voice signal under adverse noise conditions. Voice signal affected by real time noise signals is passed through amplitude estimator and from there on, it is processed in cepstral and multiband domain to extract the features of speech signal. The extracted speech is used to reconstruct the original signal which is free from real noise.