Predictive measures of directional benefit part 2: verification of different approaches to estimating directional...

In this investigation, the relation between various directivity measures and subject performance with directional microphone hearing aids was determined.

Test devices included first- and second-order directional microphones. Recordings of sentences and noise (Hearing in Noise Test, HINT) were made through each test device in simple, complex, and anisotropic background noise conditions. Twenty-six subjects, with normal hearing, were administered the HINT test recordings and directional benefit was computed. These measures were correlated to theoretical, free-field, and Knowles Electronic Manikin for Acoustic Research (KEMAR) directivity index (DI) values, as well as front-to-back ratios (FBR), in situ signal-to-noise ratios (SNR), and a newly proposed Db SNR, wherein a predictive value of the SNR improvement is calculated as a function of the noise source incidence.

The different predictive scores showed high correlation to the measured directional benefit scores in the complex (diffuse-like) background noise condition (r = 0.89 to 0.97, p < 0.05) but not across all background noise conditions (r = 0.45 to 0.97, p < 0.05). The Db SNR approach and the in situ SNR measures provided excellent prediction of subject performance in all background noise conditions (0.85 to 0.97, p < 0.05). None of the predictive measures could account for the effects of reverberation on the speech signal (r = 0.35 to 0.40, p < 0.05).

For environments that included a discrete number of noise sources, the in situ SNR and Db SNR estimates were most predictive of subject performance. No predictive approach was indicative of the directional benefit achieved when the speech was also subjected to reverberation (temporal distortion). This finding has implications for real-world estimates of directional benefit.