Target audibility in speech mixtures requires measuring glimpsed time-frequency regions, not quiet thresholds

Evidence

Accounting for audibility effects in speech mixtures is not straightforward. While it is common to measure performance in quiet for the target stimuli used, this does not incorporate the fact that portions of the target are completely masked, which greatly reduces redundancy in the speech signal. Thus a new measure of “masked target audibility” was introduced. Simple energy-based analyses have been used to quantify the available target in monaural speech mixtures (e.g. the ideal binary mask described by Wang 2005 ; ideal time-frequency segregation as explored by Brungart et al. 2006 ; and the glimpsing model of Cooke 2006 ). The basic approach is to identify regions in the time-frequency plane where the target energy exceeds the masker energy. The number of these glimpses is reduced as the SNR decreases, or as more masker talkers are added to the mixture. — Dijk, Pim van, Physiology, Psychoacoustics and…, 1753

Categorization

• DEFINE: Defines ‘masked target audibility’ and the glimpsing measurement approach in the time-frequency plane.
• corroborated: 2 independent passages support this claim — Dijk, Pim van, Physiology, Psychoacoustics and…, 1753