Because I’m gradually going deaf (I’ve lost a lot of my hearing over 2kHz already), I’ve been learning about hearing aids—I’ve even been thinking of doing “thrift-store science” with them, with a plan for making my own test box and 2cc coupler. I have not bought hearing aids yet, but will probably need to do so by next year.
So far, I’ve read two books to bring myself up to speed on modern hearing-aid design:
- Digital Hearing Aids James M. Kates [Plural Publishing 2008]
- A fairly easy read, with a good intro to signal-processing basics, but a bit dated.
- Hearing Aids, 2nd ed. Harvey Dillon [Thieme 2012]
- A much more thorough look at the acoustical physics and prescription of modern hearing aids, but surprisingly little on the signal processing or algorithms needed to make them work. The author is also way too fond of acronyms—he lists over 140 of them in the index in the back. Many of them are very similar to each other (REAG, RECD, REDD, REIG, REOG, REOIG, RESR, RETSPL, REUG, RIC, RITA, RITC, RITE, RSETS), , and he generally uses them without expanding them except the first time, which makes reading the book unnecessarily tedious.
(Actually, I’m only about 2/3 of the way through this book—I’m finding it rather dull in places.)
Both of these books are aimed at clinicians prescribing hearing aids, not engineers interested in designing or testing them, and so they have had more information than I need about how hearing aids are prescribed, and less than I would like about how they are designed and programmed (much of which is likely to be trade-secret coding).
Recently, the New York Times had an article about PSAPs (Personal Sound Amplification Products), which are hearing aids sold under a different product name, because of US regulations that require hearing aids to be certified by the FDA. That article points to research by Nicholas Reed, a Johns Hopkins audiologist, into how well these non-prescription devices work.
Note: the research article in American Journal of Medicine has Reed as the second author, with Sara Mamo as the first author—why was the female first author overlooked by NY Times? Perhaps there is another publication coming out with the authors in a different order—maybe the cited [Reed NS, Betz J, Lin FR, Mamo SK. Electroacoustic analysis of direct-to-consumer amplification devices. In preparation.]?
The NY Times article claims that the authors were impressed by three products:
- Soundhawk, which works with a smartphone (about $400) and is quite large (looks like you’re on the phone with a Bluetooth headset, and only designed to be used in one ear, I think).
- CS50+, made by Soundworld Solutions (about $350, $700 for a pair) and looking like cross between a conventional behind-the-ear model and an in-the-ear model (so combining the disadvantages of each). Customization can be done with either a computer or a smartphone (a plus for me, since I still have not bought even a dumb cell phone).Based on the specs and the documentation for their customizer app, this device appears to be a 16-channel amplifier with 16-channel noise suppression, with feedback cancellation, moderate (not adjustable) compression, and switchable omnidirectional or unidirectional mic, and the left and right ears can be separately adjusted. The American Journal of Medicine article mentions good signal-to-noise ratio, noise reduction, and speech-enhancement software.
- Bean T-coil, made by Etymotic (about $350, $600 for a pair), which looks like a conventional in-the-ear hearing aid. It is not as adjustable as the other two—they even claim “No adjustments needed; no controls to adjust” as if everyone had the same needs. According to the specs, it is a 15dB analog amplifier with wide dynamic range compression. The frequency response and compression are fixed, not adjustable. This is not a modern hearing-aid design, but one from the late 1980s! It would be a good deal at $30, but not at $350. The American Journal of Medicine article did not indicate that they were impressed with the Bean T-coil: too much low-frequency gain, only adequate signal-to-noise ratio, no noise reduction or low internal noise.
These are not low prices, but much cheaper than the medically priced ones, which average $1400 each (according to Wikipedia)—a ridiculously high price for what they contain—even the $350 ones have a huge markup, as the components probably cost in the $10–$50 range.
The Bluetooth-enabled Soundhawk and CS50+ are huge devices, which probably helps them put in all the components needed for the Bluetooth connectivity, in addition to the sound-processing components. The Companion hearing aid seems to be the same as the CS50+, but in a smaller, traditional behind-the-ear, receiver-in-the-ear aid ($450 for one, $735 for two). They claim a longer battery life, despite a smaller battery, so there are probably some compromises on the design (possibly a lower sampling rate).
You can get very cheap sound amplifiers from China ($5–$20 each), but these are often just 1-transistor amplifiers with awful distortion—equivalent to 1970s-era designs or earlier. I’ve been wondering whether it would be possible to design my own PC board to fit into a behind-the-ear case and make a hearing aid that is comparable to the Bean T-coil, but designed around my hearing losses. I don’t think that I have the patience this summer to design a full digital hearing aid like the Companion, CS50+, or Soundhawk—there is a lot of code tweaking needed to make the signal processing work right at low enough power. If I can get good answers from their customer support, I may just buy myself a pair of the Companion hearing aids.