Effects of Envelope-Vocoder Processing on F0 Discrimination and Concurrent-Vowel Identification

Abstract

The overall detrimental effects of vocoder processing are probably are due to the poor spectral representation of the lower-order harmonics. The F0 information carried in the temporal envelope is weak, susceptible to reverberation, and may not suffice for source segregation. To the extent that vocoder processing simulates cochlear implant processing, users of current implant processing schemes are unlikely to benefit from F0 differences between competing talkers when listening to speech in complex environments. The results provide further incentive for finding a way to make the information from low-order, resolved harmonics available to cochlear implant users.

Vowel enhancement effects in cochlear-implant users

“…Each envelope was used to modulate a broadband noise (50-8000 Hz), which was subsequently filtered by second-order bandpass Butterworth filters with the same center frequencies as the analysis filters, and bandwidths computed using the Greenwood (1990) function, as implemented by Bingabr et al (2008) to simulate spread of excitation similar to that typically found in CI users with monopolar stimulation. The same five vowels were used as in Qin and Oxenham (2005). The vowels were generated by producing a "baseline" harmonic tone complex with a fundamental frequency (F0) of 100 Hz.…”

Physiology, Psychoacoustics and Cognition in Normal and Impaired Hearing

“…Each indirect sound component in a reverberant space adds to the direct sound at the receiver with essentially random phase, reducing the depth of temporal-envelope modulation at the output of cochlear band-pass filters (Sabine 1922;Sayles and Winter 2008;Sayles et al 2015;Slama and Delgutte 2015). Perceptually, noise and reverberation can both decrease speech intelligibility (e.g., Nabelek 1993;Payton et al 1994), and can be particularly troublesome for cochlear-implant listeners (e.g., Qin and Oxenham 2005). However, error patterns differ substantially between noisy and reverberant spaces (Nabelek 1993;Assmann and Summerfield 2004).…”

Voice gender and the segregation of competing talkers: Perceptual learning in cochlear implant simulations

“…While several studies examined how voice gender cues are extracted by CI users and listeners attending to CI simulations (Fu et al, 2004a;Fu et al, 2005;Li and Fu, 2011;Schvartz and Chatterjee, 2012;Visram et al, 2012;Fuller et al, 2014;Gaudrain and Başkent, 2015), many questions remain about how they are used in situations where competing talker are present (Stickney et al, 2004;Qin and Oxenham, 2005;Visram et al, 2012). In the present study, experiment 1 examined the effects of voice gender on speech intelligibility in two-voice combinations where target and masker sentences were either matched in gender (MM, FF) or mismatched (MF, FM).…”