Hearing aids Hi-Fi project

[Manukatche]

Hello everybody

I'm new here and it's my first post. First, I'm French and my English is far to be perfect, so I apologize in advance if I'm hard to understanding.

I'm 40 and hard of hearing with -80 dB on mediums and a bit better in low and high frequencies, see my hearing profile and hearing aids features :





Now, I listen music with this combo :

- Smartphone Zenfone 8 pro with UAPP installed
- DAC FiiO Q3
- Symetric wire jack 2.5 mm
- IEMs Erdre Audio EA 502

I turn the DAC's volume potentiometer to 80 % with 16 ohms impedance IEMs. I need a very loud sound



I like UAPP because this app allow me to feature signal in accordance with my hearing profile by using parametric equalizer. Here is the feature :




The first picture is my favourite feature at this moment ( I didn't found the perfect trebbles feature for now )
The second picture is more clearly ( I improved weakness of my IEMs ) but corrections are positive and , of course , I ear distorsions.

This combo allow me to listen music with a very good quality ( Good quality compensate for my hearing loss ).


I'm here because I'm looking for help in order to make Hi-Fi hearing aids.

I've to found a portable device such a smartphone wich be able to capture sound, apply DSP and re-produce. Everything should take 1 ms or a bit more ( I've to avoid shift between lips reading and signal in my ears ).

This device must have a full-duplex USB port ( or 2 USB ports , I don't know what's the best feature ).

For input signal, I would like use large ( wire ) cardioid stereo microphones with a DCA. Large cardioid seems accurate in order to reproduce ear flaps, whiches can't assume their role because I wear the IEMs. These microphones will be stacked on each ear flap, the best compromise for natural capture and feedback avoidance.
I found those, but I'm a noob and I don't know if they are accurate:



For output signal, I will use a DAC with IEM's.

I want High quality audio ( 24 or 32 bits ). 64 bits would be nice but I'm not sure I'm able to ear difference between 32 and 64 bits.

I don't have a lot of knowledge about electro-acoustic and my challenge can't be take up without your help.

Which combo advice you to me ?

A big THX in advance

EDIT: By now the equalizer works for both left and right ear. It works well by reporting balance output to left but I would like to have one equalizer for each ear because they haven't the same profile.

 

[Ericglo]

I believe @René - Acculution.com worked for a hearing aid company. You might want to reach out to him for some ideas.

 

[jae]

An important thing I think will be to first make sure the headphones you are using are tuned "flat" like diffuse field or Harman iem target (without the bass boost) to ensure you are getting the most neutral sound/timbre which is important for intelligibility, then apply your custom EQ over that. It will potentially allow you to have better intelligibility at lower volumes too.

You can do that with AutoEQ (https://github.com/jaakkopasanen/AutoEq - https://github.com/jaakkopasanen/AutoEq/blob/master/results/RANKING.md), I think UAPP or one of the UAPP plugins like toneboosters might have this built in but I can't remember. I good alternative to the EQ in UAPP is using Wavelet which also has headphone profiles built in but I don't think your headphones have a profile on autoeq yet (https://play.google.com/store/apps/details?id=com.pittvandewitt.wavelet)

If you can handle deep insertion and find it comfortable enough I would suggest using a deeper insertion iem with good performance such as Etymotic. ER2SE, especially custom moulded tips/inserts will give you great noise attenuation and will probably allow you to get away with lower volumes and more favourable EQ numbers. The noise attenuation will also allow you to avoid more problems with microphone feedback if you plan on putting them near your ears. And it has the appropriate FR out of the box without needing to pre-EQ (so you will only have to EQ for your loss, not EQ for the headphone FR+your loss)

One thing to note is that even if you listen at louder volumes or correct for your loss, the distortions from your hearing loss may be non-linear. For example even at equal loudness you may find that you have more distortion in certain bands, and it may be in bands where your loss isn't as bad. This is not really something that can be found out with just a regular audiogram and will need some other speech tests or just trial and error or a good understanding on identifying FR on your part. Some people have high levels of loss but negligible distortion and some people have less hearing loss with lots of distortion, or vice versa. If you find you have high distortion in a certain band at typical listening volumes, you may have to attenuate adjacent EQ bands on either side to minimise distortion in that band. This is extremely important if those bands that you may have distortion in are important in speech. Since your mid range hearing loss is the worse, you will need to be careful with too much bass which you already seemed to have figured out (see what I said about starting with flat FR).

Modern hearing aids can seem to "take care" of some of these problems by using all sorts of interesting algorithms to handle correction, noise, feedback etc. and AI/learning derived from audiometric research and patient-reported symptoms based on their threshold and other data. Also the corrections provided for your hearing aid are very specific to that hearing aid model and would vary based on the style, dome insertion depth, dome venting etc so its not always easy to translate a provided hearing aid EQ to arbitrary iem.

I would also suggest using pkane's Earful app (https://www.audiosciencereview.com/...ds/beta-test-earful-a-hearing-test-app.14863/) which you can effectively use to do your own audiograms at home, except you can use a larger bandwidth and do as many data points as you want to get more accurate curves for your filters. Be careful with correcting frequency extremes, but it can still be useful and give you better results. If you don't want to use ER2SE for your listening setup, ER2SE will be a good reliable asset in doing your own at-home audiometric tests.

You can also consider an omnidirectional mic, perhaps? I was thinking about this recently and something like the Wireless GO II (https://www.rode.com/user-guides/wirelessgo-ii) could be an option. It comes with a mic but you can use others with it. With this you have the option of monitoring the audio directly so you can go right from the device -> headphones potentially (perhaps with some kind of DSP dac/amp in between for your correction, I believe member E1DA here makes such a product) or recorder -> phone (processing)-> headphones. It might be worth it into looking at similar portable recorders.

Per-channel corrections are obviously a huge issue. In the mean time I've contacted a few devs of software I like about this, perhaps it is something they will implement in future updates.

You can probably also simplify your setup quite a bit by ditching the large (albeit portable) headphone amp by using a good dac/amp cable-type dongle. Good dongles should really have more than enough power for most iems even at ear splitting levels.

 

[jae]

https://www.jabra.com/hearing/enhance-plus products like these are also going to start getting a lot more popular because of COVID now that the FDA are approving these off-the-shelf hearing aid/buds. Even apple I think is already adding hearing assist stuff to their products. Hopefully a matter of time before there are some really hi-fi solutions out there

 

[DanielT]

Ask your doctor for tips and advice first and talk to hearing experts. Examine (as you probably have already done) scientific reports that are published online.Trust the research, the overall knowledge base. Participate in surveys, if they are looking for test subjects.
Since I am not an expert on hearing loss and the human ear, I should not speculate on what should be a good solution. I end by saying: good luck!

Feel free to tell us about the development.

 

[Manukatche]

Hello guys

I'm still on work, I had to learn a lot of things about audiology, audio-tech, programming and electro-acoustic. I'm able to understand how I can make an earing aid Not perfectly of course I have some lacks yet
JAE, your help is very welcome, big thx to you
DanielT, here you are my fell free to tell you about the development

I will receive soon my IEMs Er2se and custom tips by ADV. I will do my audiogram then.

I didn't buy the hardware I have to fill some shadow zones But it will be a Raspberry 4 ( I hope to found this on ebay cause shortage ), hifiberry DAC/ADC and hifiberry DSP. I will use Sigma Audio to drive that.
2 omnidirectional microphones and 2 directional microphones with pre amp.
I'm looking for an amp. If I don't found that, I will follow JAE suggestion about using a DAC AMP such my fiio Q3.

Taking sometimes but I need time to learn how to make earing aids.

I will come back soon to introduce you my prototype

Nice day you guys