Ah, electric. Those electrons are really pretty dumb, and don't take suggestions as to what they are expected do very seriously, and even commands are haughtily ignored. We are thus, ashamedly, forced into slave labor. But instead of whips, chains, stun guns, stupefying injections and the like (solitary confinement proved to be self defeating), we have resorted to volts and gauss to whip them into shape.
Now electrifying an autoharp isn't exactly trivial, but is well within the capabilites on modern technology. We need to get rid of those winding pegs that tighten the strings and replace them with magnetostrictive transducers (or cheaply and crudely with a solenoid whose current can be adjusted to give the right amount of pull). We then put a piezoelectric array on the sound board so we can follow each string very precisely (array indexes are exactly like those that locate the pixel you're looking at now at the top right end of that w
A quick strum across the stings picks out all the array elements of each sting, one by one, and a fast fourier transform on our internal palm held type of computer (internal, or better in the players pocket with an simple US bus to the autoharp) tells us the frequency of each string, and the difference between that and in a stored scale is converted D to A, and a hybrid analog- digital capture-and-hold give us the voltage to send to the magnetostrictive transducer to pull the strings to proper pitch. (With simple-minded 12 tone equal temperament tuning we can bypass that table and simply calculate the proper pitch for each string, dividing down our computer's quartz cristal oscillator in the usual way to get A=440.0000000 Hz. (Ears are very forgiving of small errors so there's really little need to tie our computer to the cesium clock time-frequency standard in Colorado (It's simply an matter of local pride (one upsmanship) that they do that there, and it only takes them a short optical fiber to tie into it.)
Now, a few nanoseconds later, when they are all at proper pitch, we make another strum and fourier transform it sting by string and get the ratios of the harmonics of each string and compare with a stored table of ideal values for each string. (A simple analytical distribution approximation would probably suffice, so we wouldn't have to type up the tables.) We can then tell if someone made a mircokink in winding the strings, giving us a small region of chaos in the string. There is no cause for alarm, as we simply we do a linear least squares fit on the pixel values for the string and locate these chaoses precisely and simply drop the chaotic pixel elements out of the total electrical signal for the string. A very slight loss of information (Information =2*T(tune length in seconds)*W(frequency response range of good human ears (inadequate if your audience is dogs). Constant air pressure doesn't make any difference to the ear's response to the phonons, so we can ignore the additive 1 that sometimes appears in our equation. It's usually forgotten, anyhow.
Nothing has been said here about volume, but autoharps only rarely get up to 120 decibels, the threshold of pain. Here, that 1 we neglected in the information equation could tell us our eardrums are about to pop, but we wouldn't want to alarm the audicence with trivia like that. What they don't know won't hurt them until it's too late, anyhow.
With elaboration, we can get rid of strum thumb too, but only arthritic autoharp players have worried about this, and not to the point of funding a research institute to study the various options by a cost-benefit analysis.
I'm the old fashioned type who'd baulk at getting rid of the strings, too, so I'm not even going the hint as to how that can be done.