> have there been any significant developments in the lossless compression of audio since then?
One thing that is lacking in audio compression is the same thing as the difference between JPEG and MPEG - using past data to predict future data and only storing the difference.
Music has lots of repeating notes and passages. Isolating them (from other notes played at the same time), and only storing the slight change of how the note was played this time, vs last time should greatly increase compression ratios.
But I have not seen any audio compression that does this.
(Note this applies equally to lossless and lossy compression.)
> using past data to predict future data and only storing the difference
Actually what you've described is essentially how all of lossless audio compression really works. You pre-process the signal to make it so that the 'important' parts don't take a bunch of space to store, then you feed it to a predictor, and then encode your difference in a way that it doesn't take a bunch of space to store. You can try to tune each step to try to make your next step perform better.
Lossy compression can be made to work similar, because you can just store a less accurate difference between what you predict and the original.
The predictor looks at a 'window', which is usually defined in terms of n samples. If you make your window too short, you are not capturing any meaningful periodicity; if you make it too long you have sacrificed some error resilience and decoding convenience in exchange for hoping to get a better match for your predictor.
Speech codecs (lossy) basically operate just like you describe, see [1]
One thing that is lacking in audio compression is the same thing as the difference between JPEG and MPEG - using past data to predict future data and only storing the difference.
Music has lots of repeating notes and passages. Isolating them (from other notes played at the same time), and only storing the slight change of how the note was played this time, vs last time should greatly increase compression ratios.
But I have not seen any audio compression that does this.
(Note this applies equally to lossless and lossy compression.)