Anybody know if a left-right ear synchronization mechanism exists inside MFI & ASHA hearing aid?

Related, even though it doesn’t directly answer your question: here is a summary data sheet for an existing HA chip:

https://www.onsemi.com/pdf/datasheet/e8300-d.pdf7

This chip is likely used by Starkey in their latest Edge AI HAs.

As you found in the Bluetooth LE documents, synchronizing to a streaming source isn’t difficult since you have similar packets arriving at each side.

The thing I don’t have a solution for is synchronizing live audio. If you want to maintain simultaneous sampling between the two sides for the purpose of spatial audio (i.e., inter-aural phase), then some kind of communications scheme is needed. Perhaps ping-pong messaging where comm delay is deduced and then arrival time of the sync event in the message can be corrected. This trick is used in femtosecond (!) distributed timing systems over fiber optics. I’d be interested in finding out if/how this is done commercially.

Yes, I don’t think I can find more details about the MFI. Hope I can find the technical detail of ASHA.

A quick google search yields this regarding MFI:

https://mfi.apple.com/en/how-it-works.html5

But I’m curious, what are you trying to achieve? Is this for an OTC hearing aid?

Hi wtolkien,
Thanks for your info. I’m a technical guy, and I’m evaluating whether a lower power chip without a TWS solution but with BLE capability can be used for a Hearing Aid product. It’s not like a mature solution like Onsemi in the hearing aid industry.

The CROS situation is fairly unique, but not needed by most users. (Pretty rare set-up; not available with all versions of products and sometimes needs a dedicated sender unit)

Most hearing aids have a direct system duplex to the phone for calls or unidirectional for streaming. The proprietary aspects involve how they communicate for response matching, beam-forming and directional pairing, but stops short (I believe) of full signal transfer due to power needs.

There was a method of using NFMI to do the local comms to leave the BLE aspect free for remote comms, but I’m not sure who does this now. (Possibly Sonova, possibly everyone) Afaik, it used the same aerial but obviously pulled from different bandwidths with appropriate filtering. The advantage of this is that using the NFMI was possible with negligible latency, but if you have chop it into packets and then reassemble them there would be too much delay for face-to-face comms.
It doesn’t matter so much on the streaming set-up as you can introduce a bit more latency into the screen presentation, but with a mic that might be slightly behind the phase and a/d stage before the packet is built/unbuilt you can’t use it to build an array in live reproduction or you’ll be massively adrift with comparison of the ‘live’ signal.

@Jayden, keep in mind that HAs, if they are designed to last a full day, have a power budged of around 2mW. That’s a tall order and it’s why specialized hearing aid silicon (like the OnSemi chip) have dedicated hardware to perform common hearing aid functions in hardware.

It’s also why many over-the-counter HAs (which likely don’t use custom silicon) only last a few hours.

This company tries to spin their use of a general purpose chip as an advantage. I disagree.

https://www.hiorka.com/2

@wtolkien , I agree that power consumption is very important.

Back on the topic, I found a demo code on GitHub, and I briefly went through the code. It seems that the left and right earbuds don’t need to sync with each other. They can handle the delay or loss packets themselves, so maybe I can find some answers in this repo.
https://github.com/markusjellitsch/android_asha/tree/master2

Note that humans are sensitive to an ear-to-ear interaural time difference in the neighborhood of 10 microseconds.

10 us is about right. For direction finding via differential phase (the way we do it at high frequencies), that would be an azimuth uncertainty of for instance 18 degrees at 5 kHz. As I mentioned earlier, the biggest challenge for a pair of HAs using their microphones is inter-instrument synchronization to this level. AND all the latencies on each side have to match at that level, which is REALLY HARD when you’re talking about complex arbitrary filters! FWIW, my Phonaks have zero directionality above about 1 kHz where my natural hearing rolls off. Higher pitches appear somewhere between my ears, not out in space.

I hate to be that person and I obviously don’t know the time scale for your project, but… you understand that ASHA (and probably mfi) is going to sink beneath the waves in coming years? LE Audio should become the dominant standard. LE Audio L&R streams are supposed to be independent of each other ie no communication between buds. There’s some kind of time-based scheme where packets simply get dropped if they don’t arrive on time (and that’s the limit of my technical knowledge). Having said that I’ve experienced very rare occasions with LE Audio where the streams seem to be unsynced for a few seconds.

Hi @d_Wooluf ,
At the beginning, I have no idea about MFI and ASHA, and then I know they are the solutions to reduce the power consumption of A2DP+HFP in MFA, MFI replaces A2DP(2014)/HFP(2021), and ASHA can only replace the A2DP, the connectivity may be a problem. I do some basic research on LE audio, technically, it can provides the same effect as A2DP +HFP to customers and even better in some scenarios.

The chip I’m evaluating does not support MFI and ASHA profile, but it can support LE audio, as you know, MFI and ASHA hearing aids are very popular models, so I do some researchs on the the necessity of hearing aid manufacturers switching to LE Audio.

By now, I found some hearing aids announced they support LE Audio, but only auracast feature, I’m not sure if they have plan to switch the technical paths,maybe some have changed from ASHA to HAP/TMAP(the hearing aid profile, telephony and media audio profile), but I don’t see it.