Richardson L M, Busby P A, Blamey P J, Clark G M
Department of Otolaryngology, University of Melbourne, Victoria, Australia.
Audiology. 1998 Jul-Aug;37(4):231-45. doi: 10.3109/00206099809072977.
Prosodic information is conveyed to normally-hearing listeners by variations in acoustic fundamental frequency, amplitude envelope, and duration of speech segments. This study measured cochlear implant patients' sensitivity to these parameters in electrically coded speech. The psychophysical discrimination of electric parameters used to code prosodic information, were examined, together with prosody perception using speech processing strategies which modified the contributions of these parameters. Patients were implanted with the Cochlear Limited prosthesis and used the MPEAK speech processing strategy. In the psychophysical studies, difference limens were measured for steady-state and time-varying stimuli, of different pulse rates and pulse durations, over a series of different stimulus durations. These limens were obtained using an adaptive procedure which converged on the 50 per cent correct point. In the prosody perception studies, performance was measured for the MPEAK strategy and for strategies which modified the contributions of pulse rate and pulse duration. Data were collected for five tests of prosodic contrasts. Difference limens for steady-state pulse rates were larger at higher rates (17 per cent at 400 pulses/s) than at lower rates (6 per cent at 100 pulses/s). For some patients, limens for the time-varying pulse rates were larger than those for the steady-state pulse rates while for the other patients, the limens were similar. Difference limens for pulse duration were 0.3 dB, corresponding to 4 per cent of the dynamic range, for steady-state stimuli and doubled in size for the time-varying stimuli. Prosody perception performance was generally poorer for the modified strategies than for the MPEAK strategy, suggesting that the removal of information coded by pulse rate and pulse duration reduced the perception of prosodic contrasts.
韵律信息通过声学基频、幅度包络和语音段时长的变化传达给听力正常的听众。本研究测量了人工耳蜗植入患者对电编码语音中这些参数的敏感度。研究考察了用于编码韵律信息的电参数的心理物理学辨别能力，以及使用改变这些参数贡献的语音处理策略进行的韵律感知。患者植入了科利耳有限公司的假体，并使用MPEAK语音处理策略。在心理物理学研究中，针对一系列不同的刺激时长，测量了不同脉冲频率和脉冲时长的稳态和时变刺激的差别阈限。这些阈限是使用一种自适应程序获得的，该程序收敛于50%正确点。在韵律感知研究中，测量了MPEAK策略以及改变脉冲频率和脉冲时长贡献的策略的表现。收集了五项韵律对比测试的数据。稳态脉冲频率的差别阈限在较高频率（400脉冲/秒时为17%）比较低频率（100脉冲/秒时为6%）时更大。对于一些患者，时变脉冲频率的阈限大于稳态脉冲频率的阈限，而对于其他患者，阈限相似。稳态刺激的脉冲时长差别阈限为0.3分贝，相当于动态范围的4%，时变刺激的差别阈限大小翻倍。修改后的策略的韵律感知表现通常比MPEAK策略差，这表明去除由脉冲频率和脉冲时长编码的信息会降低韵律对比的感知。
