# Audio function for hearing impairment

I have a hearing impairment. I can't hear stuff around above 2 kHz. Is there a way to convert an input sound source (file for instance, but also a microphone) and convert it to my hearing range?

I.e., if there is a mp3 music file, it is suited for healthy humans and holds sound between 20 Hz to 20 kHz. Is there a way to convert it to a file with a sound between 20 Hz to 2 kHz?

The best analogy I can imagine is image resize – it is easy to change a 1000×1000 image to 200×200 image – we lose information on the way, but it will be good if you have a 200×200 screen.

Can it be done? How?

It can't be done.

With an image, the eye recognize the similarities of two images, one of them being the double of the other.

With sound, the ear recognize intervals : a ratio of 2 is an octava, 3/2 a fifth, and so on. Then analysing a spectrum, dividing frequencies by 10, will make the intervals completly different and you won't recognize anything. The melody won't be the same. Moreover, voyels have fixed formants. An /i/ has them at 250, 2250, 3000Hz. displacing formants will create a new voyel you won't recognize.

• I know that everything is going to be different and I still want to try that, I want to hear how this range compression feels like. how it can be done? technically... Jan 12, 2022 at 14:12
• Some Digital Audio Workstations can transpose audio files. There is Reaper. But all the frequencies are shift the same. Then if you divide the frequency by 8 (3 octava), a A2 (110Hz) will be at 13Hz : you will not ear it. Then you need a similar technology with a different frequency shift function. Unfortunately these products are rarely open-source : you can’t adapt them. The main principle is slice the song in fragments, Fourier analysis, frequency shifts, inverse Fourier transform, reassembling the fragments. But I guess phases should be treated adequately (it may be more complicated). Jan 12, 2022 at 14:49

Hearing doesn't work like that. If you are hearing impaired above 2kHz, the brain will compensate for the loss of spectrum to the best of its ability. You are likely to experience attenuated harmonics above 2kHz, which may make the sound experience a little muffled, but you will still be able to distinguish speech and music, even though some of the upper partials are not going to be detected. Previous experience of these partials however is not lost by the brain and will be remembered, so you will still be able to distinguish and classify sounds based on prior experience. Attempting to pitch-change partials or entire spectra will not improve your hearing experience.

• I am totally aware of the "autocomplete" feature of the brain. but I do want to hear how this range compression feels like. how it can be done? I do know that I am going to be disappointed... Jan 12, 2022 at 14:11
• It cannot be done.
– Mark
Jan 13, 2022 at 7:12

If you convert a voice, for example and double it's frequency then it may be in your hearing range, but will sound like Alvin and the Chipmunks. If you shift music it's is almost unrecognisable. So maybe you could live with the tonal shift for communication purposes, but anything wide band - noise, for example, would probably then mask even more. If you hearing is damaged through physical or medical trauma, your brain is the best solution because it adapts. frequency shifting is extremely limited and can often then prevent your brain processing it because a voice no longer registers as a voice.