This paper presents various Signal-to-Noise Ratio improvement techniques for EEG collected Neuro-Biomedical signals for Brainstem Speech Evoked Potentials data of Audiology. We have collected EEG data for 10 different human subjects. The de-noising techniques we developed for these Auditory Brainstem Responses are Yule-Walk multiband filters, Cascaded Yule-Walk-Comb filter, and conventional wavelet transform filters: Daubechies, Symlet, Coiflet Wavelets. After this we have designed a Translation-Invariant (TI) wavelets estimation filtering technique. In our research the idea of cascaded Yule-Walk-Comb filter is giving a considerable improvement in SNR over Yule-Walk filter; in conventional wavelets Daubechies wavelets are showing better performance than all. Ultimately, TI wavelets De-noising technique is providing us the best performance than even conventional wavelets.