Related Experiment Videos
Human Frequency Following Responses to Vocoded Speech
Saradha Ananthakrishnan1, Xin Luo, Ananthanarayan Krishnan
11Department of Audiology, Speech-Language Pathology, and Deaf Studies, Towson University, Towson, Maryland; 2Department of Speech, Language, and Hearing Sciences, Purdue University, West Lafayette, Indiana; and 3Department of Speech and Hearing Science, Arizona State University, Tempe, Arizona.
Ear and Hearing
|April 1, 2017
Summary
The number of vocoder channels and temporal envelope cutoff frequency impact neural speech processing. Increasing channels up to four improves neural representation, while higher cutoff frequencies enhance temporal envelope encoding.
Area of Science:
- Auditory Neuroscience
- Speech Processing
- Cochlear Implant Research
Background:
- Vocoders simulate cochlear implant (CI) processing in normal-hearing listeners.
- Behavioral studies explored spectral/temporal cues in vocoded speech perception.
- Neural correlates of vocoded speech perception remain underexplored.
Purpose of the Study:
- Investigate neural indices of vocoded speech perception using frequency following response (FFR).
- Examine effects of varying spectral and temporal cues on brainstem neural representation.
- Analyze neural encoding of temporal envelope periodicity (F0) and temporal fine structure (TFS).
Main Methods:
- Measured FFRs in normal-hearing adults in response to vocoded speech stimuli.
- Experiment 1: Varied number of vocoder channels (1-32) with fixed temporal envelope cutoff (500 Hz).
- Experiment 2: Varied temporal envelope cutoff (50/500 Hz) and carrier type (sine/noise) with fixed channels (8).
- Applied Fast Fourier Transform to FFRs to analyze F0 and TFS representation strength.
Main Results:
- Neural representation of temporal envelope and TFS improved with 1-4 channels, plateaued at 8-16, and decreased at 32 channels.
- Higher temporal envelope cutoff frequency (500 Hz) improved brainstem F0 representation.
- No significant effect of temporal envelope cutoff on neural representation of spectral peaks near F1.
- Carrier type (sine vs. noise) did not significantly affect brainstem F0 representation at 500 Hz cutoff.
Conclusions:
- Neural encoding of temporal envelope and TFS cues shows optimal representation with up to 4 vocoder channels.
- Enhanced temporal envelope encoding at higher cutoff frequencies reflects preserved F0 periodicity.
- Neural representation of TFS cues is not improved by higher cutoff frequencies or spectral sidebands.