No abstract

Automatic Speech Recognition (ASR) system gives better result in restricted conditions but under noisy conditions it does not perform well. The main aim of ASR research work is that a machine must recognize the entire input raw signal with 100% accuracy in real time. In the presence of noise, audiovisual features play a vital role in ASR systems. This paper summarizes various robust feature extraction techniques to study the performance of raw speech signal in automatic speech recognition. We also overview some recently proposed methods on the speech recognition, illustrating their pros and cons together with their detailed computational steps compared to other well known techniques.

No abstract

No abstract

No abstract