SSA
Thank you. Makes good sense.
The frequency shifting is somewhat unpredictable I expect because of the physical situation you describe. I hypothesize that the way to handle this is to use a methodology of repeated testing with real world words while iterating through many of the available features of an advanced HA. For example with my P70s, with 20 frequency bins, some control over the frequency shifting (ie the Sound Recover) and other features might result in a modified sound envelope that uses whatever physical recognition I have left. I know how I would do it…I need a HA with some pretty good AI to relatively consistently shift certain word sounds into a envelope that I can actually differentiate…this along with training, should allow a pretty good increase in speech recognition. I already know two things that help…cupping my ear toward the sound…which I expect makes good use of our highly evolved evolutionary ability to sustain damage and lousy hearing environments and recognize a wide variety of sounds dependably, and secondly, concentrating more on the speech to kick in higher order cognitive processing. I want an HA that helps that along…semi intelligently.
One specific way to do that now is to program a semi automatic repetivie test that makes HA changes, tracks changes to word recognition, then uses AI methodology to more quickly optimize available help from the HA. The current methods are too crude…just use one or two tests…then spread out the iteration over days weeks or months. I would speculate that about 30 minutes a day for a couple weeks might be sufficient…the allow that to soak…then repeat in a month or two. This is kinda like what happens now…but far to few test points and iteractions…and the changes may be to subtle and counterintuitive to allow real optimization by us ordinary mortals…
At any rate, by DIY fitting with the P70’s (starting from the “pro” fitting) has already resulted in some perceived improvement…I will see how it goes over the next few months. I will keep repeating the same word recognition test with the same physical controls (volume, location, spatial arrangement, etc) as I make changes over time.
Perhaps someone out there with far better programming ability than I can hack a HA fitting program to include such a testing regime?
Ethan