I'm not sure if you even need to do anything like this. The placebo effect comes about due to participant's expectation that the treatment will work. I think the placebo effect is such a big factor with tinnitus because tinnitus is related to stress. If I have a strong expectation that I am going to have significant success with the treatment in the near future, then this is likely to reduce my stress, which will actually cause the treatment to be successful (at least for a while anyway)
.
So, in the trial if you subject the participant to a battery of tests before and after treatment, pay a lot of attention to them, use a device that's obviously created for the purpose of treating tinnitus, use staff with medical qualifications, then you are likely to set this expectation. It probably doesn't matter that there is no tingling on the tongue (and staff could explain to the participant that the threshold is set so low that there will be no sensation; it would better if the staff used believed this as well). Tingling on the tongue would add to the overall effect, and would contribute to convincing the patient that they are getting the real deal, but it's probably not necessary.
The other way that you can try to reduce the placebo effect is to have several different treatment arms, and show that one arm is superior to another, and hence demonstrate a real effect. This is what Neuromod tried to do in their trial. But all 3 arms had the same results in THI reduction after 12 weeks (although they were different after 12 months, and participants with hyperacusis did experience a difference). This means that either:
a) There is something about the treatment (non-placebo) that causes it to be effective (the music played, the tongue stimulation etc) which is common to all 3 arms, or
b) The common thing for all 3 arms is the placebo effect.
I am leaning towards b).