Codecs under construction?

[Thorgal]

With the coming of sonicstage 3.3 I saw a few people giving their thoughts on the codecs and how they are compared to earlier versions of sonicstage.

I was just wondering: Are the codecs really getting better or is this just whisfull thinking. Are the codecs still being worked on? Is there any official confirmation (Sony) that this is truly the case?

Are all the codecs still being worked on, or is it only on Atrac 3 plus? I read a posting saying that LP2 sounds better than in previous versions. If this is true (and I have no evidence for that, not on the Internet, and also my ears hear no difference) than I think also LP4 is under construction, which I find hard to believe.

Other question: Will it ever be possible to reach Hi-SP quality with Si-LP?

Just curious about your thoughts on this matter.

[Sparky191]

Equally are the codecs in the hardware being improved. I'd assume for older codecs to be backward compatible they won't be improved. Only the new ones, Atrac3+ tec.

Edited November 21, 2005 by Sparky191

[ROMBUSTERS]

It'll be a cold day in hell when Hi-LP at 64kbps reaches the quality and high-dimensional sound of Hi-SP at 256kbps

[Alexx]

Equally are the codecs in the hardware being improved. I'd assume for older codecs to be backward compatible they won't be improved. Only the new ones, Atrac3+ tec.

why

If someone told you 15 years ago that a new scheme caled "atrac" would give you 1.4 Mbps sound quaity, but actually only operate @ 292 Kps you'd call them a lier

maybe in 5 years ATRAC will have refined again

However I dont think we will be storing the data on small diskettes

[KJ_Palmer]

I'm not sure you could say 292k ATRAC is really CD quality, though with Type-R encoding it is very good indeed. I'm certainly happy with 320 or 352kbps, though I think the trend may be towards higher capacity media and storage of such higher bitrates as PCM or lossless formats (like lossless ATRAC ), rather than improving lossy compression techniques...

Edited November 21, 2005 by KJ_Palmer

[dex Otaku]

I'd be willing to bet that ATRAC [sP] itself is probably still being developed, as SDDS uses it for film soundtracks, and Sony appear to be seriously committed to what is currently the only 8-channel surround scheme used in the film industry. [interestingly - I have never been able to find sufficient technical data on SDDS to figure out whether it actually uses SP or not; their literature implies it, but does not confirm it. I would hazard a guess that it's more likely SDDS uses a VBR version which allocates bandwidth by complexity by channel, similar to most uses of joint-stereo encoding, as opposed to discrete stereo encoding.]

I have serious doubts that LP2 or LP4 are being refined any further. I would doubt that any ATRAC or ATRAC3 hardware-encoders seen by consumers will see any change from the encoders that have been around since before HiMD was introduced. Type S and Type R have both been around for a while, and AFAIK no improvements have been made to either.

That said, my findings re: LP2 were found when comparing [by memory alone, thereby totally fallible] with what I remember from SS 2.1 and 2.2. I tested with the same music, and if the difference were truly minor I doubt I would have noticed and would have simply continued saying that LP2 from SS is crap, period. It did seem less like crap than before, though, and I'll stand by my opinion, which is just that.

Software codecs being improved is also a completely separate matter from the codecs being improved at any base level. I still stick with my earlier [as of mid-2004] thoughts on the fact that the software codecs for all a/3/+ bitrates in SS were compromised quality-wise in order for gains in encoding speed, to satisfy the vast majority of users who would rather have quicker response than higher quality.

Decoders are fairly easy to standardise whether hardware or software, as the basics of any given format are pretty much set in stone on that side of things. The decoder either works or it doesn't; if it's having qualtiy problems of any kind, that indicates a serious error in implementation [i.e. actually not following the format's standards], not just a compromise in coding.

Improvements in encoders can be created and implemented as long as their output follows the spec expected by the decoder. My assertion would be that, given the compromises that are plainly audible in the LP2 codec [hardware encoding sounds nothing like softwre encoding to my ears in this regard], the software codec shows further room for improvement. All they have to do is stay within the boundaries defined by the spec / decoder.

As a parallel example - MP3, in the hands of the lame group, has been tweaked sufficiently that lame's VBR encoding exceeds the quality of the reference encoder made by the company who owns MP3. I'd expect that Sony at least have the potential to increase the quality of all their codecs' encoders at least on the software front in the same manner. However - the fact that atrac/3/plus are proprietary, closed systems does imply that less work will likely be done on such, and less progress likely to be seen over a short time.

Incidentally, the quality-mode options in SS 3.3 still have no apparent effect to my ears. With a lack of any ABX system that would permit direct use of OMA files, I have no way to really test this opinion scientifically.

Oh, and - I would concur with ROMBusters that 64kbps will never achieve the relative quality of 256kbps. I still find it far less offensive than many other low-bitrate formats, but not really listenable under most circumstances.

[greenmachine]

Incidentally, the quality-mode options in SS 3.3 still have no apparent effect to my ears. With a lack of any ABX system that would permit direct use of OMA files, I have no way to really test this opinion scientifically.

How about decoding (himdrenderer) before testing? Most PC-based ABX systems can handle Windows PCM (*.wav).

[dex Otaku]

Good suggestion, gm. Perhaps I'll put my ears on for that some time soon.

[Thorgal]

As a parallel example - MP3, in the hands of the lame group, has been tweaked sufficiently that lame's VBR encoding exceeds the quality of the reference encoder made by the company who owns MP3. I'd expect that Sony at least have the potential to increase the quality of all their codecs' encoders at least on the software front in the same manner. However - the fact that atrac/3/plus are proprietary, closed systems does imply that less work will likely be done on such, and less progress likely to be seen over a short time.

Incidentally, the quality-mode options in SS 3.3 still have no apparent effect to my ears. With a lack of any ABX system that would permit direct use of OMA files, I have no way to really test this opinion scientifically.

Oh, and - I would concur with ROMBusters that 64kbps will never achieve the relative quality of 256kbps. I still find it far less offensive than many other low-bitrate formats, but not really listenable under most circumstances.

Ok, I compared the codecs to versions like 3.1 and 3.2 and I noticed not much difference. Comparing to SS 2.X is indeed another thing. Pity I don't have any transfers left done with SS 2.X

Glad to read progress is still being made. Pity though that progress is very little over time, due to the fact that atrac is proprietary. Which makes me wonder, cause Sony must also recognize this fact. They can never hold up to formats like mp3 or ogg vorbis. Doesn't this mean that in the end Sony will end the development of atrac and adapt on open but secured alternative?