Adam Mickiewicz University, Institute of Acoustics, Poznan, Umultowska 85, Poland.
Hear Res. 2010 Oct 1;269(1-2):162-8. doi: 10.1016/j.heares.2010.06.016. Epub 2010 Jul 13.
This study assessed the effects of speech misidentification and cognitive processing errors in normal-hearing adults listening to degraded auditory input signals simulating cochlear implants in reverberation conditions. Three variables were controlled: number of vocoder channels (six and twelve), instantaneous frequency change rate (none, 50, 400 Hz), and enclosures (different reverberation conditions). The analyses were made on the basis of: (a) nonsense word recognition scores for eight young normal-hearing listeners, (b) 'ease of listening' based on the time of response, and (c) the subjective measure of difficulty. The maximum score of speech intelligibility in cochlear implant simulation was 70% for non-reverberant conditions with a 12-channel vocoder and changes of instantaneous frequency limited to 400 Hz. In the presence of reflections, word misidentification was about 10-20 percentage points higher. There was little difference between the 50 and 400 Hz frequency modulation cut-off for the 12-channel vocoder; however, in the case of six channels this difference was more significant. The results of the experiment suggest that the information other than F0, that is carried by FM, can be sufficient to improve speech intelligibility in the real-world conditions.
本研究评估了在混响条件下，正常听力成年人聆听模拟人工耳蜗的退化听觉输入信号时，言语识别错误和认知处理错误的影响。控制了三个变量：声码器通道数（6 个和 12 个）、瞬时频率变化率（无、50Hz、400Hz）和外壳（不同混响条件）。分析基于：（a）8 名年轻正常听力听众的无意义单词识别分数，（b）基于响应时间的“聆听舒适度”，以及（c）主观难度测量。在非混响条件下，模拟人工耳蜗的言语可懂度最高可达 70%，使用 12 个声码器，瞬时频率变化限制在 400Hz 以内。在存在反射的情况下，单词识别错误率会高出约 10-20 个百分点。12 声道声码器的 50Hz 和 400Hz 频率调制截止频率之间几乎没有差异；但是，在 6 声道的情况下，这种差异更为显著。实验结果表明，除了 F0 之外，FM 携带的信息足以提高现实世界条件下的言语可懂度。
Zotero 插件