Evaluation of the importance of time-frequency contributions to speech intelligibility in noise

Although requiring prior knowledge makes the ideal binary mask an impractical algorithm, substantial increases in measured intelligibility make it a desirable benchmark. While this benchmark has been studied extensively, many questions remain about the factors that influence the intelligibility of binary-masked speech with non-ideal masks. To date, researchers have used primarily uniformly random, uncorrelated mask errors and independently presented error types (i.e., false positives and negatives) to characterize the influence of estimation errors on intelligibility. However, practical estimation algorithms produce masks that contain errors of both types and with non-trivial amounts of structure. This paper introduces an investigation framework for binary masks and presents listener studies that use this framework to illustrate how interactions between error types and structure affect intelligibility. First, this study demonstrates that clustering (i.e., a form of structure) of mask errors reduces intelligibility. Furthermore, while previous research has suggested that false positives are more detrimental to intelligibility than false negatives, this study indicates that false negatives can be equally detrimental to intelligibility when they contain structure or when both error types are present. Finally, this study shows that listeners tolerate fewer mask errors when both types of errors are present, especially when the errors contain structure.

Section: Introductionmentioning

confidence: 72%

Structure in time-frequency binary masking errors and its impact on speech intelligibility

Kressner

Rozell

2015

“…5 and STOI scores in Table I afford an opportunity to assess the accuracy of intelligibility prediction. STOI has been shown to be more accurate than many alternative metrics, such as the classic speech intelligibility index, and has become a standard objective speech intelligibility metric (Taal et al, 2011;Yu et al, 2014) for NH listeners. For the IEEE corpus with the same male speaker employed here, Taal et al (2011) provide parameter values of a logistic transfer function that maps STOI scores to percent-correct numbers.…”

Section: Discussionmentioning

confidence: 99%

An algorithm to increase speech intelligibility for hearing-impaired listeners in novel segments of the same noise type

Healy

Yoho

Chen

et al. 2015

104

Machine learning algorithms to segregate speech from background noise hold considerable promise for alleviating limitations associated with hearing impairment. One of the most important considerations for implementing these algorithms into devices such as hearing aids and cochlear implants involves their ability to generalize to conditions not employed during the training stage. A major challenge involves the generalization to novel noise segments. In the current study, sentences were segregated from multi-talker babble and from cafeteria noise using an algorithm that employs deep neural networks to estimate the ideal ratio mask. Importantly, the algorithm was trained on segments of noise and tested using entirely novel segments of the same nonstationary noise type. Substantial sentence-intelligibility benefit was observed for hearing-impaired listeners in both noise types, despite the use of unseen noise segments during the test stage. Interestingly, normal-hearing listeners displayed benefit in babble but not in cafeteria noise. This result highlights the importance of evaluating these algorithms not only in human subjects, but in members of the actual target population.

“…Since a higher H-FA score does not necessarily produce a higher intelligibility score, H-FA may also be unfit as a cost function for algorithm design. Yu et al (2014) tried to address some of the limitations of H-FA when they proposed the loudness-weighted H-FA, a mask-based metric that takes into account the relative importance of each error. However, the importance weights in the metric were fit to masks that employ an alternate definition of the IBM (i.e., the "target binary mask"; Kjems et al, 2009) and that use an FFT-based frequency decomposition.…”

Section: Discussionmentioning

confidence: 99%

“…Given that STOI evaluates the reconstructed signals as a whole rather than each classification decision independently, it holds promise for being able to predict outcomes for masks with different error distributions because it can take into account the perceptual relevance of the errors. Several other metrics have been proposed to assess sound source separation algorithms, such as the loudnessweighted H-FA (Yu et al, 2014), the IBM ratio (Hummersone et al, 2011), and the intelligibility metric based on an auditory preprocessing model (Christiansen et al, 2010). However, these metrics have gained limited traction due to either lacking generalizability or accessibility.…”

Section: Introductionmentioning

confidence: 99%

Outcome measures based on classification performance fail to predict the intelligibility of binary-masked speech

Kressner¹,

May²,

Rozell³

2016

To date, the most commonly used outcome measure for assessing ideal binary mask estimation algorithms is based on the difference between the hit rate and the false alarm rate (H-FA). Recently, the error distribution has been shown to substantially affect intelligibility. However, H-FA treats each mask unit independently and does not take into account how errors are distributed. Alternatively, algorithms can be evaluated with the short-time objective intelligibility (STOI) metric using the reconstructed speech. This study investigates the ability of H-FA and STOI to predict intelligibility for binary-masked speech using masks with different error distributions. The results demonstrate the inability of H-FA to predict the behavioral intelligibility and also illustrate the limitations of STOI. Since every estimation algorithm will make errors that are distributed in different ways, performance evaluations should not be made solely on the basis of these metrics.