Comparing the effects of vocoded speech on behavioural vowel recognition and the frequency following response

This study used the frequency following response (FFR) as an objective measure to understand the effects of varying spectral cues on brainstem neural representation in response to noise-vocoded speech, and describe the relationship between behavioural vowel recognition and the FFR peak F0 amplitudes.

A within-subjects design was used to compare behavioural vowel recognition using the Iowa vowel stimuli and FFRs in response to a/da/stimulus in an unprocessed condition as well as five noise-vocoded conditions with varying numbers of simulated channels (1, 2, 4, 8, and 16).

Fifteen adult listeners with normal hearing participated.

Behavioural vowel recognition performance improved significantly when the number of channels increased from 2-4, 4-8, 8-16, and 16 to unprocessed. However, no significant increases in FFR peak F0 amplitudes were seen across any condition. Additionally, no relationship was observed between F0 amplitudes and behavioural vowel recognition scores.

Improvements in noise-vocoded speech perception are driven by increases in spectral resolution (i.e., channel number), whereas FFRs are driven by temporal periodicity in vocoded signal waveforms. Consequently, FFRs may grossly capture neural phase-locking to the temporal features of vocoded signals but may not be a useful index for understanding how changes in spectral resolution impact perception.