TAK codec (Any news, seems silent for some time)
Reply #18 – 2012-09-23 22:34:57
All lossless formats are going nowhere when considering they all succeeded in outputting identical audio data. In my experience from the beginning, all lossless formats were neck-in-neck. I elaborate on this by mentioning the main aspects to handle are: compression ratio, speed(encoding/decoding) and efficiency(encoding/decoding). Usually gains of either of the first two usually sacrificed the latter. TAK proved to be an exception to what previously existed and its merits are largely indisputable. Naturally, people want more and request* their specific needs to be catered to.
In the world of digital audio, lossless audio compression is a niche at best. Just the same, I do not find using any lossless audio compressor risky. So, if ALAC takes over the world it just means I have to spend two minutes to learn its command line (for decoding ). The next breakthrough for lossless audio compression will likely take advantage of powerful processor functions as well as multi-processor capabilities. This has already begun. Another area might be mobile, but I think it would be better to have one's music library encoded with Opus in cloud storage for on-the-air streaming (I think of mobile devices with limited memory and lack of a microSD slot). I believe it has been well known for years the theoretical limitations of the ratio expected with lossless audio- particularly CD's. It seems obvious the next step forward for lossless compressors are ways to better compress heavily dynamically compressed music, since most lossless audio compressors already do well on regular organic music. Nowadays with the loud modern recordings (in addition to louder reissues of older albums) this could be an area to raise the bar. It may also involve a completely different technology that in its infant stages would run very slow. This is not how TAK works. My understanding is TAK does not employ exhaustive, brute-force methods on audio compression, but rather uses a much more elegant method by utilizing a sub-set of filters that have been well-tested and proven. (100% of the thanks goes to the author for that. ) I have many 32-bit float files that can not be negotiated by the existing version of TAK, but being I use it for all my CD images hardly makes TAK useless, irrelevant and going nowhere unless CD's stop being pressed. * euphemism