Speech Recognition Based Automated Credit Card Module

This paper presents an approach to the recognition ofspeech signal using frequency spectral informationwith Mel frequency for the improvement of speechfeature representation in a HMM based recognitionapproach. The Mel frequency approach exploits thefrequency observation for speech signal in a givenresolution which results in resolution featureoverlapping resulting in recognition limit. Resolutiondecomposition with frequency mapping approach for aHMM based speech recognition system. TheSimulation results show a improvement in the qualitymetrics of speech recognition wrt. to computationaltime, learning accuracy for a speech recognitionsystem.