Calibration of a mobile hearing aid prototype and its validation: Towards transparent listening with commodity hardware
* Presenting author
Mobile hearing aid prototypes running open-source software allow for immediate testing under field conditions, and easy distribution of algorithms. Besides low latency and acoustic feedback, an accurate calibration of the device is important for the validity of any experiments. Ideally, the listener would not perceive any difference between listening with and without the device, especially in level and tone colour (transparent listening).To test if prototypes built of commodity hardware components can be calibrated out-of-the-box, dummy head recordings in approximated diffuse field conditions were conducted for five pairs of earphones with built-in microphones. Two average FIR compensation filters were derived that equalize the microphone and the response of the headphone at the eardrum appropriately. The obtained generic calibration was validated by measuring hearing thresholds of four normal hearing subjects. The results show that the application of the generic calibration leads to about 2 dB higher variation between the hearing thresholds of the subjects than obtained by measuring the individual hearing thresholds in an open-ear, diffuse field condition (6 dB vs. 4 dB).A remaining issue is to reduce the comb-filter effect at low frequencies, potentially possible by a suitable manipulation of the gain table.