[ Contents | Next | Previous | Up ]
From: Greg
Date: 10/10/02
Time: 10:41:00 PM
Remote Name: 68.105.255.171
Hey Steve,
You raise some great points--and ones that I should have probably included in the manuscript. Self-generated (auditory) speech feedback is basically worthless during times of pre-speech inaudible silent blocks; this is where a combination of prosthetic management can really benefit from good ol’ fashioned behavioral therapy (i.e., easy phonation onsets to begin the feedback loop). Another option that is now possible via digital processing may be to introduce a digitally recorded voiced sonorant (such as /a/) prior to speech production. (e.g., in times of silent blocks, a voiced sonorant could be introduced into the user’s ear via an RF remote control; this phenomenon was found to induce fluency in those who stutter.) While masking could also be used--it may be less effective due to the binaural nature of implementation. Secondly, background noise is definitely an inherent problem in analog altered auditory feedback--and still is not “solved” with the present digital algorithms. However, every iteration in the firmware is a quantum leap from its predecessor…and the current 3.x firmware version is surprisingly good at isolating self-generated speech sounds from background noise. It’s certainly not perfect, but it’s certainly outperforming what I ever imagined. While I never imagined myself saying this--in time, the signal-to-noise problem may become a non-issue.
Anyway, thanks for the great thread … I’ve got a poster in Atlanta--Hope to see you there,
Greg