I want to find some papers that say precisely what circumstances they used to measure sound absorption due to angle, which also specifies the measured amounts along the frequency spectrum.
Let us know if you find them.
We don't regularly use HVAC in Norwegian homes, we do not have flammable gas in the average home.
HVAC ducts are only one source of leakage. There are plenty of opportunities for flanking paths, structure-born paths, penetrations for pipe or wires, etc.
I've tried to express why this information is probably not available, but you're welcome to search. The biggest problem I see with trying to develop transmission loss figures for different angles of incidence would be that the data would also have to include distance from the source to the partition, as well as the total size of the partition. This is because of the wavelengths involved. At 1KHz, the approximate wavelength is 1'. If you have a source 10' away from a 10' wide by 10' high partition, the wave hitting the wall will reverse phase every few feet along the wall. Moving the source to any angle doesn't change that, though the exact location and of the phase reversals will be somewhat different. The wall, in any case, is subject to the same type of mechanical stimulus, and assuming a wall of uniform stiffness, the amount of transmission loss through the wall will not change appreciably either. In other words, no appreciable change in transmission loss vs angle at an average of 10' from the source. If we bump that distance out to 1000', the wave front becomes nearly parallel to the wall, and changing the angle of incidence to 45 degrees would change the mechanical stimulus at the wall form a single unified wave front to several waves of varying phase, which could in theory change the measured transmission loss. Let us know if you get a 1000' by 1000' room in which to run those tests. Oh, buy the way, it should be anechoic.
If you lower the frequency, you change the number and location of wave phase reversals, until you get somewhere around the Schroeder frequency, then then there's no point in analyzing the number of phase reversals along a partition, it begins to act as a diaphragm at all frequencies below. And, by the way, the average wall is terrible for transmission loss at low frequencies.
Again, the issues that degrade transmission loss are many, and of far greater impact.
For all the above reasons, I personally would abandon the search for the data and concentrate on things that make a difference in transmission loss, like wall construction, eliminating perforations and flanking paths.