US6701291B2 - Automatic Speech Recognition with Psychoacoustically-Based Feature Extraction, Using Easily-Tunable Single-Shape Filters Along Logarithmic-Frequency Axis