Parameter tuning of time-frequency masking algorithms for reverberant artifact removal within the cochlear implant stimulus

Abstract

Cochlear implant recipients struggle to understand speech in reverberant environments. To restore speech perception, artifacts due to reverberant reflections can be removed from the cochlear implant stimulus by applying a matrix of gain values, a technique referred to as time-frequency masking. In this study, two common time-frequency masking strategies are implemented within cochlear implant processing, either introducing complete retention or deletion of stimulus components using a binary mask or continuous attenuation of stimulus components using a ratio mask. Parameters of each masking strategy control the level of attenuation imposed by the gain values. In this study, we perceptually tune the parameters of the masking strategy to determine a balance between speech retention and artifact removal. We measure the intelligibility of reverberant signals mitigated by each strategy with speech recognition testing in normal-hearing listeners using vocoding as a simulation of cochlear implant perception. For both masking strategies, we find parameterizations that maximize the intelligibility of the mitigated signals. At the best-performing parameterizations, binary-masked reverberant signals yield larger intelligibility improvements than ratio-masked signals. The results provide a perceptually optimized objective for the removal of reverberant artifacts from cochlear implant stimuli, facilitating improved speech recognition performance for cochlear implant recipients in reverberant environments.

Keywords: ACE processing; Cochlear Implants; Reverberation; Time-frequency masking.

Figure 1.

Gain function curves mapping SRR values to mask values when (A) varying τ for the IBM, (B) varying α (at β = 1) for the IRM, and (C) varying β (at α = 1) for the IRM.

Figure 2.

Electrodograms of the speech token “A boy fell from the window,” showing the CI stimulation pattern of (A) the direct path signal; (B) the reverberant signal; and mitigated reverberant signals obtained by applying (C, E, G) ideal binary masks (IBMs) and ideal ratio masks (IRMs) with increasing parameter values, τ and β, respectively, to the reverberant signal. The mask parameter controls the sparsity of the mask, resulting in under-attenuation (C and D) to over-attenuation (G and H) of the reverberant signal as the parameter value increases.

Figure 3.

Percent of phonemes correctly identified in (A) IBM and (B) IRM conditions as a function of τ or β, respectively. Each marker indicates the average score over twenty listeners, with error bars indicating the standard deviation. Performance given the direct path of the reverberant signal (DP) and the unmitigated reverberant signal (UN) are given to the left in each plot. Asterisks (*) above mask conditions indicates significant pairwise differences from the unmitigated condition.