Yes, to do this effectively requires analyzing frequency bands individually. That's the only way to maintain fidelity. It'd be best to use continuous wavelet transform. I'm guessing this cannot be done in real time. So it will be a matter of converting 2 channel files into 3 channel files in advance. Or just waiting a while for your song to start. There may be a way to speed it up a lot with minimal noticeable difference.
That seems very reminisant of the design philosophy behind the Professional Dolby SR noise reduction system for recording 🙂
Agree, and the creators of these technologies are still being awarded today meaning that all the awards given till date are still not enough for the pathbreaking changes Dolby has brought, especially to the cinema stage, besides their NR tapes. Without Dolby, we would be running mono optical audio with 5kHz bandwidth !!!
https://news.dolby.com/en-WW/237764...-cannes-marking-50-years-of-dolby-s-work-in-f
And an interesting article
https://www.film-tech.com/warehouse/manuals/dolbychronology.pdf
To be clear, DPL had steering between L,C,R and S... but the surround was a single channel. I think you are correct that DPL II added what we had called a split-surround channel... and at some point the rear surrounds got added, bringing the total speakers up from 5.1 to 7.1
Pretty much as the cinema version worked. Because of the way film sound was read off the film, using a slit of light to hit a photo receptor, it was also difficult to ensure minimal phase mis-alignment. The light source went through a slit and lens system and the slightest rotation of the slit caused timing errors (delay) betwen L and R from the projector, before even reaching the electronics of the audio processor.
VHS Hi Fi had vastly superior audio performance than any other medium at the time it was introduced, only really bettered by DAT and the ADCs which used VHS as the storage mediums.
Without more processing, when a surround signal was present the C would be eliminated naturally but L and R were still present and quite audible. Much like what happens when you reverse a single speaker in your home. Many people don't even realize it happened until they wonder what happened to the drums.
My point is there was a lot of subtle decision making that had to be done all in analog using integrators, VCAs and a complex summation before the final signal to each channel was ready.
Even when it all worked perfectly once you heard the level changes occurring you could always pick out what the matrix decoders were doing. While I miss the old days of working in theater sound I am also really happy that modern film music has progressed as much as it has and I no longer hear the DPL steering.
🙂