Hi Aaron, I'm not 100% sure I'm following, but it seems that in your FOH mixing, you're hearing lots of squealing feedback from hearing aids?? I'm assuming you're not running metal shows? How are hearing aids louder than what you're pumping out of the monitors?
Yeah, a metal show would easily drown that out. The problem comes up in much tamer things, like a folk band, or a single speaker that no longer has to yell to the back row.
For those, the PA tends to not consider acoustic feedback quite as seriously, and so it's more likely to do that than for a metal concert. Which means that even though I'm about 30dB softer than a metal concert, I'm *still* on the edge of feedback and I'm constantly listening for it. And then a hearing aid in the audience, does...
At any rate, feedback suppression is going to be a crucial part of this system (though it'll be nearly impossible to do well in this particular prototype, because of the size of the mics--MEMS mics). My plan is to design hearing aids that are nearly foolproof for users to insert and that don't follow the typical bean to in-ear-canal via thin wire design.
That would be awesome! Lofty goal, considering what already exists and still has those problems, but if you can do it, I think both the users and lots of show crews would love it!
Like having somebody stage whisper in your ear.
Yeah that'd be annoying! Though considering that they're only meant to fill in what's missing, and that's typically just the highest frequencies, it's understandable that they would do that. "Stage whisper" is just the top end and not much else, and that's exactly what most hearing aids are *meant* to do, because everything else is still there and doesn't need to be added to.
I have never seen a hearing aid designed without trying to subsitute technology for what the human brain already does...
I've mentioned in a different thread but not here yet, that I have a project on-hold for other reasons that includes what is essentially a hearing aid by a different name. Mics in a character's ears, headphones or earbuds for the performer, and DSP in between. I'm intentionally making no attempt to do anything except for corrective EQ (makes the character design a lot easier, to not have to think about its effect as an acoustic EQ), protective limiting, and a highpass filter to not add to what gets through anyway. I'm just going to measure most of that when the time comes, instead of trying to model it...
No connection at all between ears: separate stereo all the way through, as if the actor's ears are simply moved to where the character has them, and the actor's brain takes it from there.
I'm not using a Teensy for this, because its audio library is hard-coded for lesser specs than what I need, and I'm not going to rewrite the USB stack!
96kHz, solely for the converters' latency. Compared to 44.1 or 48kHz, which is more than adequate for a live stage, 96k allows them to use a more gradual filter in the converter chip itself that has fewer samples of latency, in addition to less time per sample, and thus hopefully line up with a short acoustic path around the entire system. I thought about 192k for even more of the same reason, but for what *I'm* doing, I don't think I need it *that* short, and the DSP still has to keep up! (clocks per sample...)
And of course, the DSP code takes a single sample all the way through and out before grabbing the next (no buffer, like some easy-to-use DSP chips have), just so it doesn't balloon the latency that I got rid of from the converters. If the DSP has a buffer, then there's no point at all in running faster than 48k. You're already well past lining up with a short acoustic path.
(USB is only for diagnostics and troubleshooting - what does <this point> in the chain sound like? - and as an eventual sound card for audio recording and injection. The mic-to-speaker path does not use USB at all, and so the USB buffers don't matter for that.)
