Effects of stimulation rate on speech recognition with cochlear implants
Effects of stimulation rate on speech recognition with cochlear implants
复制标题
DOI：
10.1159/000084027
复制
发表时间：
2005-01-01
影响因子：
1.6
通讯作者：
Cruz, RJ
中科院分区：
文献类型：
--
作者：
Friesen, LM;Shannon, RV;Cruz, RJ
在12名植入三种类型人工耳蜗的听者中，测量了刺激脉冲频率对音素和语音识别的影响。在5名配备有具有4或8个电极的连续交错采样（CIS）处理器的Clarion C1受试者、4名配备有具有4、8、12或16个电极的CIS处理器的Nucleus 24受试者和3名配备有具有4、8、12和16个电极的CIS处理器的Clarion C2受试者中测量辅音和元音的识别以及单词和句子的识别。根据器械、使用的电极数量和刺激策略，每个电极的刺激频率范围为200 Hz至5000 Hz以上。在接受实验处理器之前，听众也用他们原来的处理器对相同的材料进行了测试。所有测试都是在安静的收听条件下进行的，在数据收集之前基本上没有使用实验处理器的练习。使用原始处理器的听众得分最高。使用不同的刺激率的处理器，观察到听众的分数在语音理解上的差异不大。语音识别显着较差，只有在最低的刺激率和高利率，使用非交错脉冲。对于使用8、12或16个电极的处理器，语音识别是类似的。只有4电极处理器的性能明显较差。这些结果表明，目前的商业植入物的患者不能充分利用目前的语音处理器提供的频谱信息的通道数。此外，结果显示，作为刺激率的函数，性能没有显着变化，这表明高刺激率不会导致改善对语音中的时间线索的访问，至少在安静的听力条件下。版权所有（C）2005 S. Karger AG，巴塞尔。
Phoneme and speech recognition were measured as a function of stimulation pulse rate in 12 listeners with three types of cochlear implants. Identification of consonants and vowels and recognition of words and sentences were measured in 5 Clarion C1 subjects fit with continuous interleaved sampling ( CIS) processors having 4 or 8 electrodes, 4 Nucleus 24 subjects fit with CIS processors having 4, 8, 12 or 16 electrodes and 3 Clarion C2 subjects fit with CIS processors with 4, 8, 12 and 16 electrodes. Stimulation rates ranged from 200 to more than 5000 Hz per electrode, depending on the device, number of electrodes used and stimulation strategy. Listeners were also tested on the same materials with their original processor prior to receiving the experimental processors. All testing was done in quiet listening conditions with essentially no practice with the experimental processor prior to data collection. Listeners scored the highest with their original processor. Little difference in speech understanding was observed for listener scores with processors using different stimulation rates. Speech recognition was significantly poorer only at the lowest stimulation rate and at high rates that used noninterleaved pulses. Speech recognition was similar for processors using 8, 12 or 16 electrodes. Only 4- electrode processors produced a significantly poorer performance. These results suggest that patients with present commercial implants are not able to make full use of the number of channels of spectral information delivered by the present speech processors. In addition, the results show no significant change in performance as a function of stimulation rate, suggesting that high stimulation rates do not result in improved access to temporal cues in speech, at least under quiet listening conditions. Copyright (C) 2005 S. Karger AG, Basel.