Skip to content
HearingTracker
Log in

DSP microchip in Phonak?

This topic is closed.

13 replies1.7k viewsLast reply

Link to this topic

Hello,
Does anyone know which chips Phonak uses in their products to run the algorithms, for example in this one:

Audeo Lumity L50-R2

Discover Phonak Audeo Lumity hearing aids. With industry-leading innovations in speech enhancement, rechargeable design, and IP68 rating, enjoy clearer conversations in any environment. Perfect for mild to profound hearing loss.

Thanks

6 people in this conversation6 min readJump to latest

Phonak uses proprietary PRISM chips for the sound processing. It is believed the Lumity and Paradise lines share the fundamentally same version of the hardware while Lumity has newer software. PRISM is a system on a chip package with the only other major function, communications, handled by the SWORD chip which handles Roger, TV Connector, and Bluetooth, etc.

I don’t think detail design information of PRISM is available.

WH

Thanks.
Is PRISM only intended to run algorithms or is it a mixed-mode ASIC that integrates (with DSP) also AFE (analog front-end), ADC, DAC, output amplifier, etc. ?
Do you know previous Phonak concepts? Maybe in previous years they used circuits like TMS320 or ADAU1777 for running DSP ?

Don’t forget that hearing aid chips have to operate at very low powers. A type 13 battery for example gives about 1.4 volts and 280mAh and most hearing aids will operate for several days on a single type 13 cell, including Bluetooth streaming.

So I think you will find that all modern hearing aids use custom chips which are specific to hearing aid applications.

I am aware that modern hearing aid ASICs should include whole elements (i.e. AFE, ADC, DAC, etc.) for reasons of miniaturization and power consumption. For me, the two are not really the problem.
I am looking for a solution for my mother. The solution, proposed by the local distributor, is quite expensive: approximately $1700 for the basic level and approximately $8000 for the most advanced level (at this distributor there are 6 levels: 0 … 5). I thought if the basic level only offered one basic option - Frequency Shaping, I could achieve it using a $20 development board - e.g. TMS320 or even Raspberry Pi.

It will certainly include more than frequency shaping, for example it will have significant compression, unless her loss is very mild. Also it will have software to prevent feedback. And even the lowest technology level with have all manner of other processing to help with hearing speech in noise.

An alternative and more likely to be practical approach, more helpful to her, would be to buy second hand modern hearing aids for example on eBay and then fit them yourself using . But you really need to know what you are doing and there is always a risk that you will get something wrong and cause her actual harm. But I think that is less likely than if you are trying to build her hearing aids from scratch using a raspberry pi!

In reply to @Pavel_47

Not sure if you’d actually get the results your looking for like this, there’s a reason why big companies spend multiple millions every year on this very problem, so it’s the same chips running in each model, say P30 to P90, if you could find a way read/ write the chip then you could…well you get what I’m saying.
Sounds like your a type of guy, you could buy a cheap set of eBay and do a teardown and see for yourself what’s most likely used?

In reply to @tenkan

I see now that you are correct. Mention of sword disappeared with the introduction of paradise and prism. Not sure how to answer what OP is really seeking, however. Even the most basic lines in big 5 are so much more than what they have described. That is like early 2000’s stuff.

WH

I asked chatGPT what typical functions are performed in hearing aids.
Here is chatGPT answer:

1. Frequency Shaping: One of the primary tasks of the DSP module is to apply frequency shaping or equalization to the audio signal. This involves selectively amplifying or attenuating specific frequency regions to compensate for the individual's hearing loss. The amplification is typically customized based on the audiogram, which represents the individual's hearing thresholds at different frequencies.

2. Noise Reduction: DSP algorithms are employed to reduce background noise, which can negatively affect speech intelligibility. Various noise reduction techniques, such as spectral subtraction and adaptive filtering, are utilized to distinguish between speech and noise, suppressing unwanted sounds while preserving important speech cues.

3. Feedback Cancellation: Feedback occurs when sound from the hearing aid's speaker is inadvertently picked up by the microphone and re-amplified, resulting in a high-pitched whistling sound. DSP algorithms analyze the audio input to detect and cancel feedback signals in real-time, preventing or minimizing feedback occurrences.

4. Directional Microphones: Many hearing aids utilize directional microphones, which can focus on sounds coming from the front while reducing sounds from other directions. DSP algorithms are used to analyze the microphone signals and steer the directional characteristics, improving speech understanding in noisy environments.

5. Speech Enhancement: DSP techniques are employed to enhance the audibility and clarity of speech. These algorithms may emphasize specific speech features, such as formants or consonant cues, to improve speech intelligibility, particularly in challenging listening conditions.

6. Adaptive Signal Processing: Hearing aids with adaptive signal processing can continuously monitor and adjust their settings based on the acoustic environment. DSP algorithms analyze the incoming audio, detect changes in sound characteristics, and dynamically adapt the amplification, noise reduction, and other processing parameters to optimize the listening experience.

Any comments ?
Did it miss something ?

Maybe you can consider it part of #1, but compression and then there is also frequency shifting (usually lower) are functions often also provided by hearing aids. It is more than amplification to thresholds. The dynamic range is made to fit in the range of hearing available to the patient.

WH

I think in 1 the generality of answers is given away when it suggests that frequency shaping involves …”attenuating specific frequency regions”. It’s true for frequency shaping in general, but I doubt that you would ever find that in a hearing aid.

But the list is a good start. Don’t forget the hearing aid companies spend millions of $ researching this stuff though. It’s not just a matter of clever programming of an FPGA.

The book Digital Hearing Aids by James M. Kates gives an engineering-level description, dated 2008, of hearing aid processing. I got a really cheap copy a few years ago, but now it’s selling at textbook prices.

In reply to @david.hendon

@david.hendon said:

An alternative and more likely to be practical approach, more helpful to her, would be to buy second hand modern hearing aids for example on eBay and then fit them yourself using DIY. But you really need to know what you are doing and there is always a risk that you will get something wrong and cause her actual harm. But I think that is less likely than if you are trying to build her hearing aids from scratch using a raspberry pi!

I second .
In my experience having the time and the willingness to learn the software is way more important than having the latest hardware.

Most audiologists while well meaning just don’t have the time to devote to an individual patient.

Also programming hearing aids, and trying different stuff can be fun! I have mine dialed in so well, I don’t have any reason to mess with them anymore, which means I just lost an activity of interest. :grinning:

Related discussions