Different licenses exist for different classes of service--e.g., a nonprofit LPAM is a different license than LPAM commercial--and then there are those public service type broadcasts like are done by a city, for tourists, or by wildlife refuges to inform the tourist of upcoming and current activities, and such.
But NONE of those services are on shortwave bands. Strictly commercial MWBC band.
Different licenses exist for different classes of service--e.g., a nonprofit LPAM is a different license than LPAM commercial--and then there are those public service type broadcasts like are done by a city, for tourists, or by wildlife refuges to inform the tourist of upcoming and current activities, and such.
But NONE of those services are on shortwave bands. Strictly commercial MWBC band.
What about longwave (groundwave losses would be less severe, although the antenna would be much shorter relative to the wavelength) bands? Anything that might be possibly used there?
Also, is there any difference between when a non-profit individual/organization asks for public donation to cover operating costs (most NP stations I have heard do this), vs. footing the ENTIRE bill him/her/themselves (the way I'd do it)?
An experimental license from the FCC would cost $50. (It's free for a non-profit group). However the purpose of the license is to experiment and test equipment for possible new radio services. You have to show technical merit and how it would make the best use of spectrum. From your posts it sounds like you basically need to listen to the audio yourself. I don't think it likely to be issued an experimental license as it's going to be hard to justify taking up a frequency for the purpose of relaying a program to one person. It just doesn't make efficient use of the public airwaves. The FCC website has the experimental license application available for download if you want to try it.
Otherwise I can't think of any license that might be available to you as there are currently no new broadcast applications being accepted. Other services wouldn't allow you to use them for what you have in mind.
There are a few problems with me using a cell phone to do it.
1. It costs me 10c/minute on T-Mobile's prepaid service. (and I wouldn't want to tie up someone else's phone)
2. the audio quality isn't good enough. There are too many compression artifacts, and I'm not confident in its ability to pass frequencies above 4.5kHz or below 27 Hz (the minimum range I want).
If your main concern with cell service is fidelity (and cost were no object), commercial FM stations have done remote broadcasts with cell phone "combiners" which can take two independent cell phones and double up on the bandwidth for near FM audio quality. Back at the studio, another combiner re-combines the signal using two land lines. As far as the minutes are concerned, you can get other plans with a higher amount of fixed minutes. However I am sure that this equipment is not cheap. Also you mentioned STL. Again, the big boys with deep pockets go out and buy a Marti microwave link. These are all probably beyond your budget and the realm of Part 15, and makes the telco loop sound more attractive.
The "STL" would only need to go maybe 50 feet. Basically it would take the audio from inside the building, and transfer it to outside the building where the main transmitter/antenna would be (for better signal output / less building attenuation). I would think a wireless FM mic would work for the very short distances involved, even if it happened to be stepping on a local FM station (and I also would think that with the very low output level of the STL it wouldn't interfere with that FM station off the property where the station is).
Slightly off topic compared to the original intent of the question on this thread, but... for a 24/7 AM broadcast operation somewhere at or above 1500kHz with something like a properly loaded 8-foot vertical radiator on a metal (is this legal?) mast and several ground radials (I don't have room for 1/4-wave btw - they might average something like 1/16 wave), and hopefully a decent homebrew (I can't even come CLOSE (several decimal places) to affording a rangemaster) 100mW transmitter... would it be too unreasonable to expect the contour referenced in FCC Part 73.88 to reasonably cover my back yard? (btw the entire property is 90 feet high (north/south) by about 270 feet wide (east/west), about 1/2 acre. Also, how do I figure out (taking into account ground conductivity and (if possible) terrain) approximately how far lesser grade contours would be away from the transmitter? And, if having that contour in 73.88 cover my entire 1/2 acre property isn't attainable legally, how far should I expect it to go with a proper setup? Btw this one would be proposed to be in the vicinity of 32°45'40"N 116°56'50"W. Proposed frequency choices include 1500 (but it's used by an unmodulating TIS about 8 miles west), 1570 (seems to be fairly open, but there's a program I like to DX on that frequency once a week for 1/2 hour, and I'd prefer to be on 24/7 (although I could go off the air then), and it would interfere with my reception of Radio Disney 1580 from Phoenix), 1610 (there's a TIS about 10 miles to the west also), or 1680 (co-channel to Fresno's Radio Disney which is receivable here at night). Also there's a local talking house currently on 1620 (was on 1610 until fairly recently) at approx 32°45"32.08"N 116°56'43.16"W, which can barely be heard with the internal loopstick on a portable at my location.
A directional antenna would be nice (mainly to aim the main signal lobe from slightly ccw of due west to slightly ccw of due north), but pending a decent signal in those directions, or impossibility of getting a directional system set up, a ND would be ok.
Assuming non-directional signal... would it be unreasonable to expect the following basic levels of coverage? (go to google maps, and plug in the transmitter coordinates (they're approximate, but close), and click hybrid view. zoom levels start at z1 for zooming in the closest that google will let you zoom.)
z1: full receiver noise quieting (even on terrible quality tuners), S9+80dB, whatever that might be
z2: basically same as z2 to a normal consumer, but of course slightly weaker (it's about a mile to Avocado Blvd on the far left, and I've heard many part 15 stations claim about a mile of coverage - this I've always assumed to be something like an 60-90dB S/N ratio on a cheap portable. Although if it's really 20dB under the noise on a high grade communications receiver with a full wavelength longwire antenna AND 8-foot tuned loop antenna, that would be quite disappointing. :()
z3: some lesser quality radios starting to get a little bit of noise
z4: most radios have significant static, but station still receivable on most radios. GE Superradio might still have something listenable, but would have some static too. most walkmans the signal might be 6-10dB SNR
z5: If you can hear it on a portable (walkman, for example) with only the built-in antenna, you are very lucky. Also getting quite weak on most car radios.
z6: requires something like Select-A-Tenna, but with it signal is somewhat listenable
z7: even with SAT, is very faint. larger diameter loop antenna recommended - 2 or 4 foot for example
z8: basically no trace even with SAT unless ionosphere is cooperating. some static on 2-foot loop, but copyable.
z9: basically rely on critical hours or nighttime only for any significant coverage if you're using a SAT. Internal antenna only - forget it if you're a normal listener
z10: a bit weaker than z9. basically nearly impossible daytime even with longwire and 10 foot loop at receiving end
z11: maybe I'm getting my hopes too high here? (if there WAS any signal here, it'd probably even at night be all but buried under the atmospheric noise by now).