Introduction: The Frequency Shifting Hearing Aid
From EE122A, Mixed-Signal Circuits Laboratory, at Stanford University
Taught by Professor Greg Kovacs, Fall 2013
The problem:
High Frequency hearing loss (HFHL) is an inability or reduced ability to hear high frequency tones. It can originate from exposure to loud noises or simply as a result of age-related hearing degeneration. HFHL can result in reduced appreciation of music in milder cases or reduced speech comprehension in more severe cases.
An electronic hearing aid can help compensate for HFHL. However, simply amplifying high frequencies might not be the best course of action. Research indicates that amplifying high frequency tones is useless if the hearing loss is profound enough, and that attempting amplification may actually be detrimental to speech comprehension.
In some cases for more severely impaired listeners, increasing the audibility of high-frequency speech information resulted in no further improvement in speech recognition, or even decreases in speech recognition (Hogan, Turner, '98)
The idea:
As such, this project impements frequency shifting instead of amplification. This project is targeted at increasing speech recognition for users with severe HFHL (loss of sensitivity after ~3KHz). High frequency information from 3-5 kHz is shifted to 1-3 kHz and added back to the original signal.
In its current form, the project is not perfect. Imperfect carrier suppression in our modulator IC leads to a small 2 kHz tone which is distracting when used for processing speech playback. However, the system does provide noticeably clearer audio for music playback, where a low background tone is less noticeable. Regardless, this Instructable will serve as a useful primer for anyone interested in practicing with basic signal processing, analog filters, and circuit design.
Notes:
For this project, you should have a basic understanding op-amp usage, circuit schematics, representing signals in the frequency domain, and frequency filtering. As far as parts are concerned, you'll need 4-8 op-amps, an amplitude modulator IC (like the AD633), a respectable variety of resistors and capacitors, and male/female audio jacks.
Step 1: Understand the Block Diagram
Here is a high level view of how the circuit works, in a helpful block diagram.
First, the bandpass filter isolates the high-frequency information we want to shift - specifically, the 3 kHz to 5 kHz band.
Second, the modulator shifts - with a 2 kHz carrier wave - that frequency band down to 1-3 kHz. This shifting is done using the upper side-band of the negative carrier frequency. Theoretically, the carrier wave at +2 kHz should corrupt this sideband. However, using a modulation IC that promises carrier suppression (such as the AD633) should mitigate this concern.
The 4 kHz low pass filter attenuates the upper sideband of the positive carrier.
The 3 kHz low pass filter for the original signal is mostly for testing purposes. Listening to this output allows us to simulate what HFHL above 3 KHz would sound like. Also, in the event that the user has only partial hearing loss from 3-5 kHz, this filter would prevent duplication of the information in the output signal.
The summing amplifier layers the original signal with the high frequency shifted components.
NB: Depending on the modulator used, some amplification will be needed between the band pass filter and the modulator. The AD633 attenuates signals by 1/10 while modulating. The sideband is also naturally attenuated by 1/2 as a byproduct of amplitude modulation. As such, we used a 20x amplifier between the two stages.
Step 2: Understand the Theory
This sequence of images is designed to clarify the signal processing going on in the circuit. Note how each step maps to a portion of the block diagram.
Step 3: Design the Filters
At this point, the project is mostly comprised of design of the three filters. The easiest way to do this is to use TI's free WEBENCH Filter Design software located here:
http://www.ti.com/lsds/ti/analog/webench/webench-filters.page
By following the prompts, any active filter can be designed quickly and easily. Recall that our filters are:
1 Band-pass filter
Center frequency = 4 kHz
Passband width = 2 kHz
1 Low-pass filter
Cutoff frequency = 3 kHz
1 Low-pass filter
Cutoff frequency = 4 kHz
For filter implementation, we elected to use 4th order Chebyshev 1db response type with Sallen-Key topology. Chebyshev is recommended for its sharp rolloff - important for the tight frequency bands needed for this application (Consider the second low-pass filter. We have a cutoff frequency of 4 KHz and need to completely eliminate the frequency components at 6 KHz.). However, this area is a design space where the implementer might try different things for different results!
The attached images depict example filters. We used LT1056 op-amps, but any general purpose op-amp should work fine.
Step 4: Implement Amplitude Modulation
For the modulation step, we decided to use the populator AD633 analog multiplier. The AD633 is inexpensive and widely available. It also allows for carrier suppression when modulating. To set up the AD633 for amplitude modulation, use the "Linear Amplitude Modulator" circuit from the Applications Information section of the datasheet. Connect the output of the bandpass filter to the Modulation Input pins (pins 1 and 2) and connect a 2 KHz carrier wave (at approximately 1Vpp) to the Carrier Input (pin 3).
Important notes:
Between the inherent 1/10 attenuation of the AD633 and 1/2 attenuation of the sidebands during modulation, the signal we are shifting will be attenuated by 1/20. Remember to amplify the output from the bandpass filter by 20x before inputting it to the modulator. A simple non-inverting amplifier will do.
At the same time, note that the AD633 has some input voltage restrictions that are poorly defined. Input signals above about 5Vpp result in some distortion at the output. Signals below about 100mV are attenuated to near-noise at the output. We found workable voltage levels to be at about 1Vpp for the carrier wave and about 2V average for the modulation input (amplified bandpass output).
Connect pin 6, the summing input, to ground instead of the carrier signal to implement carrier suppression.
One imperfection we noted is that this carrier suppresion is not perfect, leading to an undesired 2 kHz tone on the output. The implementer should experiment with improving this!
Step 5: Implement Summation
With the three filters, the amplitude modulator, and the 20x non-inverting amplifier designed, the final step is to connect the blocks together (as per the block diagram) and route the output of the low-pass filters into a summing amplifier.
Step 6: Listen to the Output
Congratulations! At this point, you should have a working rudimentary high-frequency hearing loss frequency shifting hearing aid (HFHLFSHA).
Now, you'll want to solder leads to a male audio jack and a female audio jack (pictured on the title page).
Connect the male audio jac to an audio source (iPod, smartphone, PC, etc.), and connect the leads to the input of your circuit. Using the female audio jacks, connect a pair of headphones (or an earbud) to the output of the 3 kHz low pass filter, and another to the output of the summing amplifier. The output of the summing amplifier should be noticeably crisper, especially for music with high frequency tones (snare drums, clapping, female vocals, etc.).

Participated in the
Hardware Hacking
10 Comments
9 years ago on Introduction
I wonder if such a system could be combined with the tactile hearing aid concept-- in essence, taking high frequency signals that are beyond the users range of hearing- shifting them down into a much lower frequency and driving some kind of tactile vibrator which could be mounted somewhere on the head where the skin is sensitive to vibration. Just musing. Thanks for this instructable!
Reply 9 years ago on Introduction
High frequencies may not be detectable by skin contact, at least that is the case with me. Might sound a bit like having a dentist drilling inside my head :P
Reply 9 years ago on Introduction
Especially if you are at the dentist! But yea, fair point, for sure. As I mentioned, shifting those high frequencies down to much lower frequencies would mean that if the skin was vibrating, then high frequencies (generally) were present. In this way, if an "S" or "T" were spoken, the listener would get used to the presence of additional "data" from the skin. Eventually, one might discern a difference between the feel of an S and that of an F.
Reply 9 years ago on Introduction
This is an exciting device, one I have not seen before. I an very interest in hearing or experiencing its output. Thank you for the instructable. I Might have a go someday. I will definitely study it a bit more.
3 years ago on Introduction
I have dispcussis which means I hear a different note in each ear. A tuning fork sounds a minor third lower in my left ear than the right. Also the octaves are compressed so low notes sound sharp and high ones sound flat. I really need a way to filter sound in one ear so I hear the same pitch. I had to give up music because of this. Huey Lewis of Huey Lewis and the News has quit his professional career from the same problem. This might be a platform from which more research on this problem could be done.
5 years ago
Does anyone know where I can find a software version of this or an App that can be used on an Android Smartphone?
5 years ago
I want to make it but need a contiguous, continuously connected circuit diagram.
Or have I missed something here?
7 years ago
Many years ago in the early 70's I underwent trials with the R.N.I.D. in London using what they called a 'transposer' aid. This transposed higher freqs down to a lower pitch so that my hi-freq deafness might be helped. Unfortunately it did not work for me, but the idea seems to be the same as this device.
9 years ago on Introduction
very cool
9 years ago on Introduction
Hey, I love your project!
You see my dad is deaf in one ear, which cannot be fixed since his nerves are damaged.
BUT
For those people deaf in one ear because of an eardrum, they can still use bone induction to hear from their ear.
The way humans tell direction from sound is 1) stereo and 2) doppler effect. Sound actually changes frequency (very slightly) when it hits one ear versus another. Maybe you could design a hearing aid that not only can hear higher frequencies, but also can function if you are deaf in one ear!