AVRs for 25 years have been upscaling 5.1 to get more channels. As you know, prologic took the common audio between L&R and made a centre dialog channel, with a crossover for bass and a difference channel for surround. So people went and bought an AVR. A few years later, Prologic 2 added some steering for improved channel separation, and used 90 degree phase shifting to split the rear surrounds. Go buy a new AVR.
After separation, the L&R channels still have other sounds that are common to both channels, basically anything that was directly above or behind the actor. L&R can therefore be fed through another DPL2 decoder, where the common audio is removed, increasing channel separation, creating additional channels L wide and R wide. This is DPL EX. Go buy a new AVR.
DPL2x went one step further and used the common audio to make a front height CHANNEL. Go buy a new AVR. DPL2z went further and used the difference information to create "front presence" channels, basically front surround. Go buy a new AVR.
Logic 7 did the same thing but on the rear surround channels, looking for common audio and feeding it to a back surround. Go buy a new AVR.
The key theme here is looking for audio that appears in 2 channels, extracting it, and feeding it to a new speaker placed between the other two speakers. These sounds were not deliberately "encoded" into the mix. Sound engineers did not create many different mixes for DPL, DPL2, EX, 2X, 2Z, Logic 7 and dolby digital. Sound is mixed, and often the mix results in two channels receiving the same sound at the same volume and phase, and this sound can be extracted like it is for the centre channel.
All of this was done in analog. Then digital came along and gave the 5 channels discrete information. This made those analog upscaling AVRs redundant.
In 2013 "atmos" format appeared, and jaws dropped. Suddenly audio was "3D". The claim was that "new" digital height channels had been added to the new format, so you better go buy a new AVR. Again. But they are not "new". There are no sounds in the height speakers that are not already coming through the bed layer speakers. Sounds that are picked up equally by the LF and LS microphones will be at the same volume and phase, and can therefore be used to make sound half way between the LF and LS speakers. The left height speaker is at that location. This makes sense, because if you have a helicopter in the air on the left, you will hear it from the LF and LS speakers. Whether you have 5.1 or atmos, you will hear the helicopter in those speakers. All atmos did, was replicate the same technique used by 2x and Logic 7, but instead of applying it L to R, they did it front to back. Basically, you can use a DPL2 decoder to split the LF and LS channels to create new channels. Same for the right side. And the front, and the rear. Any modern AVR can do this.
The claims by Dolby Atmos and DTS that these are "new, discrete channels" is misleading. There are no sounds unique to the height speakers. As evidence of this, you can look at the output graph for any atmos movie, and you never see a height speaker outputting by itself. It will always replicate a bed layer speaker. They claim to be able to put a sound anywhere in 3 dimensional space, which I might agree with if you had 8 speakers making a perfect box shape. "it's xyz coordinate metadata, dude" isn't a lie, but when you only have 2 ceiling speakers, it is impossible to place a sound in 3 dimensional space. Which means that it reverts from object based to channel based. These atmos cultists are shouting "but it's got electrolytes".
In addition, when there is a height sound such as a helicopter, rain, thunder, plane, rocket, spaceship...these sounds do not come through the height speakers alone. They come through many speakers, which destroys the effect. Then there's music, which is always excessively loud out of the height speakers.
Manufacturers have decided to cease making bluray players, because physical media is dead. It's all streamed nowadays. When you stream, you have a bandwidth limit. So it becomes difficult to transmit extra channels of audio plus metadata. The result is compression, which means that atmos sounds worse than 5.1. It's not even necessary to transmit the extra information anyway, since it is always a duplication of existing channels.
So my plan is to matrix the height channels to simulate atmos, and ensure that height sounds only go to height speakers. A helicopter for example, has an infrasound frequency below 20hz. It's often a square wave, with harmonics depending on the number of rotor blades. Thunder also has infrasound, but jagged. Rain is always white noise played through all channels equally, so that's easy. Bird tweets are 2k-3k sine waves. A plane is similar to a helicopter, but I've not looked at the waveform. You don't even need FFT for these. You can use a LC circuit tuned to 1hz on the LFE channel. If voltage high, output the matrix to heights, because it's either a helicopter or thunder. If it was just bass, it would be a sine wave.
If LF and LS are in phase, matrix and output to the LH speaker.