Find the sound that keeps your tinnitus quieter for longer
After many people hear a sound near their tinnitus pitch, the tinnitus goes quieter for a while once the sound stops. Afterquiet measures that quiet spell after every sound it plays, and uses the result to choose the next sound to try.
See how it worksGet in touchThe quiet after the sound
Residual inhibition is the temporary suppression of tinnitus after a sound is switched off. Audiologists have measured it for decades. The usual test plays noise for 30 to 60 seconds, about 10 dB above the level that just covers the tinnitus, then times how long the tinnitus takes to come back.1
For most people who get it, the quiet lasts seconds to a minute. Sometimes it lasts longer.2 Not everyone gets it, and it is temporary.
What makes it useful is that you can notice it yourself, right after the sound stops. That turns "does this sound help?" into a number you can measure every single time.
Masking
Residual inhibition
The best sound is different for everyone
I have tinnitus, and my first attempt at a tool was a masking and pitch-matching app. It did not work well. My pitch match moved around from day to day, and it was easy to land an octave off.
Research points somewhere more useful than a single pitch. In two McMaster University groups, 90 people in total, the tinnitus spectrum and the depth and length of residual inhibition both followed the region where hearing was worst.8
So Afterquiet starts from your hearing and from how similar sounds at different frequencies are to your tinnitus. Then it searches inside that region, using your own measured quiet as the guide.
A loop that learns from every try
Each try is called a trial. You rate how loud your tinnitus is, listen to a sound, then keep a slider matched to your tinnitus as it comes back. The program scores the dip and picks the next sound. Then you rest and go again.
A trial takes 2 to 5 minutes including the rest, so a session holds roughly 10 to 20 of them. Every trial teaches the search something, including the ones that do nothing.
The settings of each sound stay hidden until the trial is over, so expectations do not steer the ratings.
Before the first trial: four short setup steps
One trial, start to finish
About one trial in seven is a control: silence, or a sound far from your tinnitus. They look like any other trial. They show how much of an "effect" is just expectation or the normal ups and downs of tinnitus.
What you do during a trial
When the sound stops, the slider starts at the loudness you gave at the beginning. You move it down to where the tinnitus is now, and keep it matched as the tinnitus comes back. Arrow keys work too, so you can keep your eyes closed.
The trial ends by itself once the slider has been back at your starting level for a few seconds.
Deeper and longer both count
From each curve the program takes three numbers. Depth is how far the loudness dropped. Duration is how long until it was back to 90 % of where it started. The area between the starting level and the curve combines both, so it is the main score.
The area is divided by your starting loudness. That way a trial on a loud day and a trial on a quiet day can be compared. Staying at half your usual loudness for 60 seconds would score 30.
Three kinds of sound, many settings
Afterquiet plays pure tones, band noise (a slice of noise around one frequency), and either of these with a slow pulse in loudness, called amplitude modulation. Every sound fades in and out over 20 milliseconds, so there are no clicks.
Pure tone
Band noise
Modulated noise
| Setting | Range | Notes |
|---|---|---|
| Centre frequency | 250 Hz to 16 kHz | Searched inside the range your spectrum points to. |
| Bandwidth | 1/12 to 1 octave | Noise only. A tone is treated as zero bandwidth. |
| Level | 0 to 20 dB above your masking level | Always under a hard ceiling (see Safety). |
| Duration | 30 to 90 s | Starts at 60 s, the usual clinical length. |
| Ear | left, right or both | Fixed at first, can be searched later. |
| Modulation | 2 to 40 Hz, 0 to 100 % deep | Off at first, switched on once a good region is found. |
How it chooses the next sound
There are far too many possible sounds to try them all, and each try takes minutes. So Afterquiet uses Bayesian optimization, a method built for exactly this: learning from very few, noisy measurements.
After each trial it updates a best guess of the score for every possible sound, together with how unsure it is. It then picks the sound with the best chance of beating the best one so far. Sometimes that is right next to the current best. Sometimes it is a region it has not checked yet.
After the first 4 trials
After 12 trials
These two charts were computed by the real program on a simulated listener, over centre frequency only, so they can be drawn in two dimensions. The real search covers all the settings at once.
Does the search beat guessing?
Before trying it on myself, I ran it against simulated listeners whose best sound I chose in advance. After 25 trials, the sound Afterquiet recommended scored 32.6 on average. Picking sounds at random scored 25.1. The best possible was about 39.6.
About one trial in five repeats the current best sound, because a result that does not repeat is not a result. The level is capped whatever the search wants, since louder sounds generally give more quiet and a search left alone would just turn the volume up.
The simulated listeners use an invented response pattern. This shows the machinery works. It says nothing yet about real tinnitus.
Score of the recommended sound, 8 simulated listeners
Five checks before any sound plays
These are enforced by the program itself, not just shown on screen. Every sound has to pass all of them.
Afterquiet is a personal experiment, not a medical device or medical advice. It does not treat tinnitus, and residual inhibition is temporary.
See a doctor or audiologist first if your tinnitus is new, in one ear only, pulses with your heartbeat, or came with sudden hearing loss or dizziness. Some of those need prompt care.
What it can and cannot do
It can
- Measure your own residual inhibition, the same way every time.
- Search for the sound that gives you the deepest and longest quiet.
- Tell real effects from expectation, using control trials and repeats.
- Keep every sound under hard safety limits.
It cannot
- Make tinnitus go away. The quiet is temporary.
- Work for everyone. Not everyone gets residual inhibition.
- Replace a hearing assessment. Its hearing check is a rough guide, not a clinical test.
- Promise lasting change. Whether regular use helps over weeks is an open question.
Why not cancel tinnitus with anti-phase sound?
Most tinnitus is not a sound in the ear. There is no sound wave in the ear canal to cancel. It is activity in the hearing pathway itself. Sound played into the ear goes through the inner ear and can only add activity. Also, above about 4 kHz, where most tinnitus sits, the hearing nerve barely follows the phase of a sound, so there is no phase to flip.
A commercial "phase-out" device tried it anyway. An open study without a control group reported lower loudness.3 A follow-up by the same Antwerp group found no significant cancelling, and for people with tonal tinnitus, loudness and annoyance went up.4
The exception is rare objective tinnitus from spontaneous otoacoustic emissions, where there is a real sound a sensitive in-ear microphone can pick up.
Measured before, but not used to search
| What | What it does | What it does not do |
|---|---|---|
| McMaster HNP Lab tools | Computer tools by Larry Roberts' lab that measure the tinnitus spectrum and the depth and length of residual inhibition. Free for research use.5 | They measure. They do not search for one person's best sound. |
| Quiescence | Desktop software built on residual inhibition that showed early promise. | Discontinued.2 |
| My Practical Tinnitus | A free iPhone app with a residual inhibition test, next to pitch matching and sound therapy guidance.6 | One feature among many, with no search. |
| VA and Duke hearing aid study (2023) | 20 adults streamed a personally fitted residual inhibition sound from a phone app to hearing aids for two months. Questionnaire scores improved and stayed better. Measured loudness did not change significantly.7 | The sound was set once at the clinic, then fixed. |
From laptop to pocket, if it works
Right now Afterquiet is a working prototype on a laptop with wired earphones. Next I test it on my own tinnitus over several weeks, and I will report the results here, including if it does not work.
Open questions
- Should amplitude modulation be searched from the start, or only once a good region is found?
- How to handle tinnitus that is louder on some days than others. Time of day is recorded with every trial for this.
- Does a short daily routine with the best sound change anything over weeks? The VA and Duke study hints yes for distress, not for loudness.
If you work on tinnitus and see something wrong here, or you want to know when Afterquiet is ready to try, write to me at peter@portlandaiworks.com.
Sources
- Clinical residual inhibition test: a 30 to 60 s sound at 10 dB above the minimum masking level, then stopped. Described in Quinn et al. 2023, citing Henry and Meikle (2000).
- Hearing Loss Help, Residual inhibition can temporarily eliminate tinnitus (also on Quiescence).
- Vermeire K, Heyndrickx K, De Ridder D, Van de Heyning P. Phase-shift tinnitus treatment: an open prospective clinical trial. B-ENT 2007;3 Suppl 7:65-9.
- Meeus O, Heyndrickx K, Lambrechts P, De Ridder D, Van de Heyning P. Phase-shift treatment for tinnitus of cochlear origin. Eur Arch Otorhinolaryngol 2010;267(6):881-8. Trial record NCT00833950.
- McMaster HNP Lab, tinnitus measurement software licensed.
- My Practical Tinnitus, App Store.
- Quinn CM, Vachhani JJ, Thielman EJ, Kulinski D, Sonstroem A, Henry JA, Smith SL. A pilot study to evaluate a residual inhibition technique in hearing aids for suppression of tinnitus. Semin Hear 2023;45(1):123-140.
- Roberts LE, Moffat G, Baumann M, Ward LM, Bosnyak DJ. Residual inhibition functions overlap tinnitus spectra and the region of auditory threshold shift. JARO 2008;9:417-435.
- Neff P, Michels J, Meyer M, Schecklmann M, Langguth B, Schlee W. 10 Hz amplitude modulated sounds induce short-term tinnitus suppression. Front Aging Neurosci 2017;9:130.
- Schoisswohl S, Arnds J, Schecklmann M, Langguth B, Schlee W, Neff P. Amplitude modulated noise for tinnitus suppression in tonal and noise-like tinnitus. Audiol Neurotol 2019;24(6):309-321.