Frequency compression in bimodal CI users (Sharma et al., 2021)

Purpose: Speech understanding in noise and horizontal sound localization is poor in most cochlear implant (CI) users with a hearing aid (bimodal stimulation). This study investigated the effect of static and less-extreme adaptive frequency compression in hearing aids on spatial hearing. By means of frequency compression, we aimed to restore high-frequency audibility, and thus improve sound localization and spatial speech recognition.Method: Sound-detection thresholds, sound localization, and spatial speech recognition were measured in eight bimodal CI users, with and without frequency compression. We tested two compression algorithms: a static algorithm, which compressed frequencies beyond the compression knee point (160 or 480 Hz), and an adaptive algorithm, which aimed to compress only consonants leaving vowels unaffected (adaptive knee-point frequencies from 736 to 2946 Hz).Results: Compression yielded a strong audibility benefit (high-frequency thresholds improved by 40 and 24 dB for static and adaptive compression, respectively), no meaningful improvement in localization performance (errors remained > 30 deg), and spatial speech recognition across all participants. Localization biases without compression (toward the hearing-aid and implant side for low- and high-frequency sounds, respectively) disappeared or reversed with compression. The audibility benefits provided to each bimodal user partially explained any individual improvements in localization performance; shifts in bias; and, for six out of eight participants, benefits in spatial speech recognition.Conclusions: We speculate that limiting factors such as a persistent hearing asymmetry and mismatch in spectral overlap prevent compression in bimodal users from improving sound localization. Therefore, the benefit in spatial release from masking by compression is likely due to a shift of attention to the ear with the better signal-to-noise ratio facilitated by compression, rather than an improved spatial selectivity. Supplemental Material S1. Individual audibility benefit (dB) for static and adaptive frequency compression and for different sound types. LP = low-pass; HP = high-pass; BB = broadband. Supplemental Material S2. Individual RMSE (deg) for sound localization with (static, adaptive) and without (no) frequency compression and for different sound types. LP = low-pass; HP = high-pass; BB = broadband. Supplemental Material S3. Speech recognition thresholds (dB) for S0N0 condition with and without frequency compression and the corresponding spatial release from masking [SRM (dB)] for each participant. Sharma, S., Nogueira, W., van Opstal, A. J., Chalupper, J., Mens, L. H. M., & van Wanrooij, M. M. (2021). Amount of frequency compression in bimodal cochlear implant users is a poor predictor for audibility and spatial hearing. Journal of Speech, Language, and Hearing Research. Advance online publication. https://doi.org/10.1044/2021_JSLHR-20-00653**研究目的**：多数佩戴助听器的双模式刺激使用者（即同时使用人工耳蜗（cochlear implant, CI）与助听器的使用者）在噪声环境下的言语理解能力与水平声定位能力均较差。本研究探讨了助听器中静态及程度更温和的自适应频率压缩（frequency compression）对空间听觉（spatial hearing）的影响。通过频率压缩技术，我们旨在恢复高频可听度，进而改善声音定位（sound localization）与空间言语识别（spatial speech recognition）能力。 **研究方法**：招募8名双模式CI使用者，分别在启用与未启用频率压缩的条件下，测量其声音检测阈值（sound-detection thresholds）、声音定位能力及空间言语识别表现。本次测试采用两种压缩算法：其一为静态算法，可对超过压缩拐点（compression knee point，160 Hz或480 Hz）的频率进行压缩；其二为自适应算法（adaptive algorithm），仅压缩辅音频段，保留元音不受影响（自适应拐点频率范围为736 Hz至2946 Hz）。 **研究结果**：频率压缩可显著提升高频可听度（静态压缩与自适应压缩分别使高频阈值改善40 dB与24 dB），但未对定位表现产生实质性改善（定位误差仍大于30°），且未在全体受试者中实现空间言语识别能力的提升。未启用压缩时的定位偏差表现为：低频声音偏向助听器侧，高频声音偏向人工耳蜗侧；启用压缩后，该偏差现象消失或发生反转。每位双模式使用者获得的可听度提升，可部分解释其定位表现、偏差偏移的个体改善情况；对于8名受试者中的6名，还可解释其空间言语识别能力的提升。 **研究结论**：我们推测，诸如持续性听力不对称、频谱重叠不匹配等限制因素，使得双模式使用者无法通过频率压缩获得声音定位能力的提升。因此，压缩技术带来的掩蔽空间释放（spatial release from masking, SRM）效应，更可能源于受试者的注意力向压缩后信噪比更优的耳侧转移，而非空间选择性的提升。 **补充材料** S1：针对静态与自适应频率压缩以及不同声音类型的个体可听度提升值（dB）。LP=低通滤波（low-pass, LP）；HP=高通滤波（high-pass, HP）；BB=宽带（broadband, BB）声音。 S2：针对不同声音类型，在启用（静态、自适应）与未启用（无压缩）频率压缩的条件下，声音定位的个体均方根误差（root mean square error, RMSE）（°）。LP=低通滤波；HP=高通滤波；BB=宽带声音。 S3：针对每位受试者，在启用与未启用频率压缩条件下的S0N0工况言语识别阈值（dB），以及对应的掩蔽空间释放值[spatial release from masking, SRM（dB）]。 **参考文献**：Sharma, S., Nogueira, W., van Opstal, A. J., Chalupper, J., Mens, L. H. M., & van Wanrooij, M. M. (2021). 双模式人工耳蜗使用者的频率压缩量对可听度与空间听觉的预测效果不佳。《言语、语言与听力研究杂志》，预印在线发表。https://doi.org/10.1044/2021_JSLHR-20-00653