Hi all,
Getting a bit school yard like here..
@piggywig is technically correct that not all microphones have the same sensitivities across the audible bandwidth. This can be demonstrated on an oscilloscope.
But the question remains whether or not the differences that can be indicated on an ossciloscope are significant to either the human ear, or the accuracy of app itself. Generally, with the exception of some public phone booths (receiver doused in beer and coke), it is usually possible to recognise the voice on the other side.
In fact, I make the case that a phone company whose phones doesn't allow the user/receiver to distinguish the voice of the individual on the other end of the line would struggle to stay in business
Though the sample frequencies that are pertinent to the algorithm may just happen to fall within a narrow band that an inferior microphone just happens to omit, it would be very very unlikely, as the tapering off/attentuation of sensitivity is very gradual and then only on both the very upper and lower ranges.
Personally I believe that the far more significant audio obstacle is the harmonic variance that arises from the many different patients that are presenting their coughs in the trials. Unless we have genetic clones, of the same age, diet etc; we are talking about a very significant variable that, to date, hasn't seemed to hinder the algorithm from excelling in previous trials.
I also believe there is a possible chance that we might never see ResApp on the app store, either in the near future or ever at all- instead it's function might be embedded within the telehealth companies's online user interface- you might not even see the ResApp logo. Such a scenario could allow commercial exposure to desktop PC's, tablets etc... basically anything with a microphone and a web browser.
I am sure though PiggyWig, that if one particular brand of phone is consistently faring poorer than others in the trials it will be immediately noticed.
I have a strong hunch it won't be a big concern.
Cheers
CT
