Hacker News Reader

Comment by ampdepolymerase | Hacker News Reader

For a few dollars more it is possible to put a FPGA with perhaps a neural accelerator on it. It may tremendously simplify some of the DSP stack. More importantly, there should be a full-proof app or UI to allow the user to tune the hearing aid appropriately (or coupled with a training course for audiologists in the targeted regions to do so).

View on HN

Hearing aids are tremendously constrained with regards to power and mechanics. There is no way you will be able to get an FPGA in a hearing aid. An FPGA runnning complex DSP algorithms will:

1. Get hot. I don't want a +60C FPGA in my ear.

2. Eat up the battery. The batteries are small and fiddly to replace for an 85 year old with bad eyes and arthritis in the fingers. They should last for at least a week before needing replacing. No they are generally not rechargeable in the device. That would require a connector, which takes up precious volume. Wireless recharging requires a coil with the same problem.

3. Take up way too much PCB real estate. They are just too big to fit in your ear. Sure, you can run a wire to a box outside the ear, but that messes with the fashion statement. Most hearing aid users want small and discrete, so the world isn't reminded they have a disability.

Besides, why would you put an FPGA in there? Hearing aids already feature some of the most high-grade low-power DSP ASICs available, custom made for that companys DSP algorithms.

> a connector, which take up precious volume. Wireless recharging requires a coil with the same problem.

Couldn’t a “connector” just be a tiny gold contact to meet another in a charging cradle/case like a smart watch or wireless earbuds? Would that be too big?

More than just a connector, it also needs a charging circuit, which would also take up space.

Does that charging circuit need to be inside the hearing aid, and not in the charging cradle/case/dock thing?

Generally, yes. The circuitry for power does more than charge the battery, it provides power regulation to prevent over discharge and reduce the likelihood of short circuits (among other things)

About the precious volume and heat - the technology still seems very unimpressive given we have wireless earbuds in stores. Mounting something behind the ear doesn't seem very extravagant these days and vendors only profit from this forced conservative approach because it smothers any potential competition that thinks it has to be this way.

Yes. Plenty of people prefer behind the ear devices because they are either rechargable, or brightly coloured. Young people in particular benefit from being able to chose an aid that's brightly coloured, and you see some YP modify their aids with skins or stickers.

Here's the colour range for one modern aid: https://www.fmhearingsystems.co.uk/PDF/FocusColours.jpg

Rechargable: https://www.phonak.com/uk/en/about-us/rechargeable-technolog...

Having a brightly coloured hearing aid as an adult doesn’t look professional. Are these marketed towards younger people?

As an adult, I want a brightly colored hearing aid. I don't care about hiding my disability—in fact I'd rather it be difficult for others to ignore—and I want it to be easy to find if my kids knock one out of my ears which happens less now that they're six but still happens.

Why not? People wear brightly colored jewelry or “statement” eyewear in a professional context all the time.

Are you saying that it’s “professional” to hide your disability if you have hearing loss? If so, then I strongly disagree. It’s a personal choice.

Indeed, to make an even more direct comparison: brightly coloured glasses are very common these days. They seem to be treated like neckties: it's acceptable for them to standout even in very formal situations.

I didn’t mean it to be about ‘hiding’ your disability. I wear glasses and I would never wear brightly coloured rims in a professional environment. By my own standards, in a work place you stand out by the merits of your work, not on what you’re wearing.

Dude... chill.

Even if you can be “right” you’ll be alone on Merit Island, let it go it doesn’t matter have some fun and let others have their “things” too.

This is your opinion. In many parts of the world, criticizing a piece of accessibility equipment as unprofessional is likely an HR violation, and probably grounds for termination.

It is and you are correct. It’s not unreasonable for someone to want a more neutral colour if that is their preference though. Some HA users are self conscious and hide that they are wearing them.

There are trade-offs.

Behind the ear is great when you still have reasonable hearing in a good chunk of the spectrum, so the dome from the BTE provides the boost that is additive to the natural signal.

Wireless earbuds' counterparts in the hearing aid world are called "in ear" or "in canal" or "completely in canal" or "hidden", and they provide some benefits, but do have a couple of drawbacks: they can often give the impression of a plugged ear because they keep the natural signal out (exacerbating tinnitus, which is common with hearing loss), and they can irritate the ear canal from all the friction (that part of our body is very sensitive). They can also prevent the natural outflow of cerumen (wax) and cause its accumulation.

is it possible to use the users phone for those calculations ?

Think about a dumb hearing aid, with a microphone and bluetooth connection, which is able to connect to any compatible phone and phone does the processing.

Realtime audio is tricky without hardware acceleration.

A recent smartphone has enough hardware for that, I would think, and if it hasn’t, but the idea made sense, it would be added. If the hardware can power a GPU and a million-pixel display, it can power quite a beefy DSP, too.

I would worry about the latency of a round-trip over Bluetooth. Google search tells me it’s, optimistically, 100ms single-trip, so 200ms round trip (https://www.headphonesty.com/2020/07/fix-sound-delay-bluetoo...)

Can anyone who knows about Bluetooth comment on why I can ping google in 50ms (wirelessly, even), but a Bluetooth round trip over a few centimetres takes 2 - 3X that?

Please don’t think of me as an expert, because the sum of my knowledge comes from browsing Wikipedia articles. But the summary of the issue is that Bluetooth gets all its energy efficiency and robustness by constantly scanning the spectrum it is to live in, and hopping around extremely fast to different slices of channels, essentially taking advantage of the tiny “gaps” in other signaling protocols that share the same free spectrum. In other words it operates on sloppy seconds/scraps like a vulture, which means it’s timing is very loose and cannot provide any guarantees. But it’s very low power as a result because it almost never has to shout over anyone else.

As callalex mentioned, the spread spectrum could introduce some delays into something like first packet response but this isn't necessarily the crux of the issue. Once a channel is being used, it can often stick around on that particular channel for a while so subsequent packets wouldn't necessarily have to do the whole spread spectrum dance until the devices start experiencing issues. IIRC there's some amount of stickiness to the spread spectrum utilization, but I could be wrong.

You're a tad bit incorrect on Bluetooth round trip a few centimers being 150+ms. Plenty of game controllers operate on or on top of Bluetooth and do not have 150ms latency. Headset Profile (HSP) bluetooth also does not (usually) have 150ms+ latency, usually more along the lines of maybe 30ms latency while many Bluetooth controllers will have <10ms latency.

https://www.reddit.com/r/RocketLeague/comments/8wvf9u/contro...

I'm not the person who designs codecs, but I'd wager most of the lag seen with Bluetooth audio systems come from the codecs. Bluetooth devices are often incredibly power constrained, so codecs operating over Bluetooth are usually trying to balance quality with computational complexity and dealing with packetization of the audio. When you're listening to an audio stream, it appears like a continuous stream of data. In reality, its chunks of data. You're listening to a previous chunk of data while the device is waiting to get the next packet, apply error correction to it, decode the very highly compressed data, and have the waveform ready to go out to your ears all on usually pretty cheap and highly power constrained equipment. It then has to be a bit tolerant to the above mentioned spread spectrum and a potential dropped packet here and there while still seeming to just be a continuous audio stream.

So in short, there's latency in a lot of Bluetooth audio codecs because the designers of the codecs felt the latency tradeoff was acceptable given the bandwidth, compute, power, and quality envelopes based on technology available at the time of crafting the codec.

Meanwhile, you're sending out pings on a device with keys on the keyboard bigger than the battery on a lot of bluetooth headsets with many orders of magnitude more processing power whose wireless connection is not normally that much of a power concern encoding/decoding extremely basic data to one of the companies with the most privately owned fiber with billions of dollars of computing equipment.

See madspindel's comment above ^^^ about Bluetooth 5.2 LE Audio and this:

LE Audio: Announced in January 2020 at CES by the Bluetooth SIG, LE Audio will run on the Bluetooth Low Energy radio lowering battery consumption, and allow the protocol to carry sound and add features such as one set of headphones connecting to multiple audio sources or multiple headphones connecting to one source It uses a new LC3 codec. BLE Audio will also add support for hearing aids.

And wireless introduces significant latency! You don't want wireless anything if you can at all avoid it with real time audio.

Easier said than done. I developed a hearing aid app for iOS.

full-proof -> fool-proof

NOT being pedantic, just trying to be helpful / educational (esp given the large % of non-native English speakers on HN)

Sorry the auto correct changed the spelling. I meant fool-proof. I am a native speaker.

Reply on news.ycombinator.com