Comparison between the Macro and Micro Mechanical Model using Values of the Human Cochlea vs. Fluid Mechanical Model for Automatic Speech Recognition

This paper shows a comparison between the macro and micro mechanical model, proposed by Neely and Kim, and extended by Elliot and Ku vs. mechanical fluid model proposed by Lesser and Berkeley both used in ASR tasks. These models are used to set the central frequencies of a bank filter to obtain parameters from the speech in a similar form as MFCC (Mel Frequency Cepstrum Coefficients) has been constructed. Also we show an equation that show the relation distance vs. frequency extracted from the solution of the mechanical fluid model mentioned above. Neely’s model was used with a set of different parameters of the cochlea, used by Nelly, Elliot and Ku in their works, such as mass, damping and stiffness; among others. The performance obtained was of the 98 to 100% using this model while a 98.5% performance was reached using the second propose; for a task that uses isolated digits pronounced by 5 different speakers in the Spanish language. Finally corpus SUSAS with neutral sound records with some advantages in comparison with MFCC was used.