>The frequency response of a healthy ear isn't flat across all audible frequencies, so you'd need to reference normal hearing...
Isn't the normal frequency response of a healthy ear dependent on the shape of the ear cartilage and the configuration of the ear canal and the ear drum? It would be different for every individual. Kinda like how Mom could always hear everything we did and said after bedtime while Dad, without even using his selective hearing, wouldn't even know we were still awake.
>Are you talking about basically using active noise canceling to silence tinitus?
Yes. Model the tinnitus and design the inverse filter based on the bandwidth and inject that inverse filter to become an active subtraction of the tinnitus response. I know it probably isn't possible because the noise is variable and originates in the brain instead of external to the ear so it is not easily quantifiable therefore the inverse operator will not be exact, optimum, or anything else. However, if you can model the signal then you should be able to design the inverse operator. Since the signal is just a band-limited input there is no reason why you can't dink around until you have a close enough model to be able to design the inverse filter which you would then inject as an external input thru an earpiece or some other sound generator.
I'm a geophysicist with hearing problems, not an audiologist or otolaryngologist. It sounds reasonable to me. We deal with convolution/deconvolution and other signal processing as a regular part of the job process.