(An) adaptive digital hearing aid chip with on chip human factors consideration

Abstract

A fully integrated human-like audio processor is proposed and implemented with physiological real time internal gain verification algorithm. The proposed gain verification algorithm uses a closed loop feedback mechanism to estimate an additional internal gain to a vent, a leakage path and a residual ear canal volume when a hearing aid is worn. By using the proposed algorithm a resonance gain error between the real ear canal and the test cavity can be minimized. In addition, additional gains due to the various internal signal paths also achieved with only internal iteration processes due to the closed loop feedback algorithm. To prove this algorithm with the co-operation of the hearing aid, a hearing aid chip implementation is also accomplished. The fabricated processor also gives a multi-mode function. The hearing aid mode offers a conventional hearing aid operation with accurate external sound amplification and transmission according to the user’s demands. The smart earphone mode enables to receive an electrical sound signal from an external audio source such as MP3 player. Finally the direction perception mode enables a discrimination of a sound source location which is essential to enhance an understanding of speech. The fabricated hearing aid processor introduces the possibilities of hearing aid as a personal audio assistant device to allow everyone to control and enhance the sounds around us. To reduce power dissipation and achieve high flexibility, a multi-threshold preamplifier and an adaptive-SNR analog front end is adopted A low power 16 channel DSP with interpolated finite impulse response (IFIR) filterbank for use in digital hearing aids is developed. Coefficient approximation and subsequent sub-expression elimination techniques are used to reduce the computational power requirement for each filter. Additionally, delay elements are implemented using circular buffers to minimize power dissipation from shifting of samples within the filters....