I'm surprised to see such a new standard for FLAC - the protocol itself has been in use for decades, but this IETF standard has just been published. I would love to hear about the context of this publication.
Not sure which version they're referring to here but from Xiph updates...
> The latest draft of the FLAC specification by the CELLAR working group of the IETF. This more formal specification improves on the format description you can find further down this page. It provides a better explanation of concepts like wasted bits, the implications on subframe bit-depth of using stereo decorrelation, explanantion of the actual symbols used in rice coding, inclusion of various decoding, etc
So it's just likely a long process on a path for formal definition, that was only started relatively recently in the format's history.
Further, from scanning some forum/list discussions from a few years ago, the thinking or motivation seems to be benefits for implementers having docs to make building a FLAC decoder easier. Which indirectly benefits users with good implementations. And then for long term storage/archival having the standard defined in an open/official document helps ensure it won't fade into oblivion. Or succumb to implementations that are incompatible with future hardware etc. Protected by the IETF as it were. I suppose MP3 went thru similar steps many moons ago.
Currently archives often store audio uncompressed, in broadcast wave format. This results in much more e-waste, shelf space required etc. But they can't switch to just any lossless codec, they need one that can still be decoded 50+ years in the future.
The old specification document didn't specify everything, you needed to read the source code or compare with a working implementation. Will people still understand the programming languages current implementations are written in then? We don't know. So this document should be stand-alone, so one could implement a decoder from scratch (without any other reference) in the future.
With this document, archives can more confidently use FLAC to compress audio and thus work more efficiently.
I was curious if instead of a dedicated lossless codec you could just encode with Opus plus the (compressed) bit difference from the original data. Google sent me to this informative HN thread: https://news.ycombinator.com/item?id=14598929
Opus (or other lossy codecs) try to minimize the perceptible loss. The phase can be completely different, so compressing the difference results in the same size as compressing the original... But you'll need to store the opus bit too.
There are several such codecs however, with a lossy and a lossless part. Not because it is easier, but for compatibility. DTS HD Master Audio is one of them: it has a backwards compatible DTS core.
