SS01: NOISE REDUCTION IN HEARINGS AIDS AND ITS EVALUATION

HEARING AID NOISE REDUCTION BASED ON SII-OPTIMIZATION – THEORETICAL FOUNDATIONS AND CLINICAL RESULTS

Karolina Smeds

ORCA Europe, Stockholm, Sweden.

Maximizing speech intelligibility is a fundamental underlying rationale for gain prescription, and the rationale has been implemented for instance in NAL-NL1, a well-known prescription for nonlinear hearing aids. The NAL-NL1 prescription is based on optimization of the speech intelligibility index (SII) in quiet for an average speech spectrum.

Modern hearing aids normally incorporate single-microphone noise reduction algorithms. Using multi-channel adaptive amplitude compression, the gain in each channel may be determined not only by the total signal level, but also by the speech audibility in that channel. Frequency-specific gain adaptations are then applied to reduce the gain in any time segment and any frequency band where the speech audibility is temporarily low. However, measurements at our laboratory have shown that noise reduction algorithms from various hearing aid manufacturers work in very different ways, and they are probably based on very different underlying rationales.

The noise reduction system implemented in a recent hearing aid modifies a prescribed, baseline, hearing aid gain using SII-optimizations. These optimizations are made based on the hearing loss of the hearing aid wearer and on short-term estimates of the speech and noise spectra. The gain can be modified relative to the baseline prescribed gain, and the algorithm tends to lead to large gain reductions in the low-frequency range, and to some increased gain at high frequencies for certain signal-to-noise ratios.

In the current presentation, an SII-based rationale for the noise reduction system implemented in a recent hearing aid will be presented in addition to clinical results where the particular noise reduction system is evaluated in the field and in the laboratory.

karolina.smeds@orca-eu.info