Data from: Development of sound localization strategies in children with bilateral cochlear implants

Stimulus WAVE Files used in localization taskStimuli consisted of bi-syllabic Children’s Spondees that have been previously used for studies on spatial unmasking of speech in children with NH [14], and with BiCIs[15], recorded with a male voice at a sampling rate of 44 kHz. The speech corpus had 25 words. On each trial one of the words was randomly chosen from the list, with replacement. Thus, on any given trial the stimulus was unpredictable, thereby reducing the availability of reliable spectral cues that might be used in the computation of sound source location. Stimuli were balanced for loudness using equal root-mean-square levels for all words. Stimuli were amplified and sent to the loudspeakers via Tucker Davis Technologies System III hardware. Stimulus levels averaged 60 dB SPL, with a random variation between 56 and 64 dB SPL (±4 dB) from trial to trial to minimize the availability of overall level cues.Stimulus Files.zipMatLab code used to calculate and plot Localization SensitivityAnalyses of localization sensitivity employed a statistical non-parametric Kruskal-Wallis test to examine the relative independence of responses to each stimulus location, given a single response to an unknown stimulus. For each pair of locations in the loudspeaker array, the two response groups were analyzed using pairwise localization sensitivity index (LSI) values based on the z-scores from the p-value obtained bythe non-parametric Kruskal-Wallis test. The analysis produces a matrix of LSIs in coordinates of response locations. For N=15 target locations, the total number of LSI values was 105 [15*(15-1)/2=105]. The matrix is symmetrical about the diagonal. The matrixwas interpolated via the MATLAB contour function to find a contour of LSI with a z score of 1.65 (the 95th confidence level as the statistical threshold indicative of two locations being perceptually differentiated). We defined the contour of LSI = 1.65 as the sensitivity threshold. The overall localization sensitivity (LS) was characterized for individuals by averaging LSIs in the LSI matrix.analysis.txt本研究定位任务中使用的WAVE格式刺激音频文件： 刺激材料为双音节儿童斯邦迪词（Children’s Spondees），此前已被应用于针对正常听力（Normal Hearing, NH）儿童与双侧人工耳蜗（Bilateral Cochlear Implants, BiCIs）使用者的言语空间去掩蔽相关研究[14,15]，由男声录制，采样率为44 kHz。该语音语料库共包含25个词汇。每次实验试次均从词汇列表中随机抽取一个词汇（允许重复抽取），因此任意试次中的刺激均不可预测，由此降低了可用于声源定位计算的可靠频谱线索的可利用性。所有词汇的均方根（root-mean-square, RMS）声压水平保持一致，以实现刺激响度的平衡。刺激信号经放大后，通过塔克·戴维斯技术（Tucker Davis Technologies）System III硬件设备传输至扬声器。刺激声的平均声压级为60 dB SPL，各试次的声压级在56~64 dB SPL（±4 dB）范围内随机波动，以尽可能降低整体强度线索的可利用性。 Stimulus Files.zip 用于计算并绘制定位敏感度的MatLab代码： 定位敏感度分析采用克鲁斯卡尔-沃利斯（Kruskal-Wallis）非参数统计检验，旨在基于未知刺激的单次响应，检验不同刺激声源位置下的响应相对独立性。对于扬声器阵列中的每一对声源位置，基于该非参数克鲁斯卡尔-沃利斯检验所得p值对应的z分数，通过配对定位敏感度指数（Localization Sensitivity Index, LSI）对两组响应开展分析。该分析将生成以响应位置为坐标的LSI矩阵。当目标声源位置数量N=15时，LSI值的总数量为105[15×(15-1)/2=105]，且该矩阵关于主对角线对称。通过MATLAB的等高线函数对该矩阵进行插值，以获取LSI值为1.65的等高线（z分数1.65对应95%置信水平，可作为判断两个声源位置在感知上可区分的统计阈值）。本研究将LSI=1.65的等高线定义为定位敏感度阈值。个体的整体定位敏感度（Localization Sensitivity, LS）通过对LSI矩阵内所有LSI值求取平均来表征。 analysis.txt