After following the discussion in the 128 kbps listening test thread I wanted to check what Sony actually claims. I stumbled on this page:
http://www.sony.net/Products/ATRAC3/tech/aal.html

I have never heard about the format before. The part that describes extraction of ATRAC3 or ATRAC3plus data from lossless ATRAC files is especially interesting.

If it can include a small lossy version in the standard ATRAC3(+) format without increasing the lossless size and can separate a lossy file without delay, it would be something not seen before with other formats. For example, Wavepack Hybrid is quite different because it has two file componets and the lossy files are big and not usable with portable players.


Edit: removed my guess and edited a bit. I think Sony really claims what I wrote above.

It's indeed interesting, but I can't see any mention on: "without increasing the lossless size". [edit: you only guessed it sorry]
WavPack hybrid is very thrifty (wv + wvc are only 5...12 kbps bigger than pure lossless from my experience). For Atrac lossless it has to be seen.

If Sony is cheating an ATRAC lossless file includes separate lossless and lossy parts. E.g. the lossless part is 20 MB and the lossy part is 4 MB, which makes a 24 MB file.

If they are not cheating the file size is only 20 MB or only slightly more and Sony has developed an effective pealing system that makes identical files with separately encoded ATRAC3(+) files.

Yes, in this case it would be cheating.
Keep in mind that atrac3plus is supposed to perform excellently at 48 kbps only (see this). A lossless + 48 kbps lossy file wouldn't decrease that much the compression ratio (48 kbps = less than 4% = approx. difference between flac and MAC -high). That's why it would be better to wait and experiment this technology. Sony is a champion for playing with words. The Net-MD was supposed to be compatible with MP3 (it means that MP3 are reencoded before transfer)

I believe it really is true lossless this time. They have Hi-MD units out now that record in Linear PCM and their ATRAC3plus format isn't very bad as a standalone. Lossless is the only step to go next...

I am think ing of buying a Sony NWA3000. Does anyone know if it's compatible with VBR mp3s?

I haven't owned a portable player before and am concerned that I might find playback of VBR tracks a problem.

Thanks

If it doesn't playback VBR files correctly, it is not totally mp3 compliant.

Wow, if it's really what it claims to be, then Sony has found some way to effectively peel the lossless filem like Alex said.

@adam917
I don't think it's about whether it's real lossless or not, it's more about whether the lossless format is what Sony promised it to be.
Sony claimed that from the lossless format, it can extract the ATRAC3 or ATRACplus data out of it.
Which means it probably uses some kind of a bit peeling technique (if it's really like that, that is).

It's different from WavPack's Hybrid or OptimFROG's DualStream, where it consists of two files, the lossy format, and its correction file.


Is this the right thread to ask?
Anyway, almost all modern MP3 players should be able to play VBR MP3s w/ no problem.

The newly published MPEG-4 SLS allows effective peeling of lossless layers from the lossless file. The lossless is somewhat very close to the best lossless codec in market, and the lossy format can be in AAC format. If your SLS file was ripped from a CD (with no DRM), then the lossy AAC will be playable on iPods as well. Obviously no devices supports SLS yet, since most people had not even heard of this new codec...

Atrac Advance "Useless" is the better term. Why? Let's say you rip a track to Atrac Advance lossless + 64kbps as the lossy part to save space. Then let's say you want to get a 256kbps version from that track. You would expect that Sonicstage would transcode the lossless part to 256kbps, right? Wrong! It trascodes the 64kbps(!) to 256kbps. Any transcoding involved use the lossy part, making the lossless part useless. So you have to rip to Atrac Advance lossless + highest available lossy bitrate to minimize any reduction in quality due transcoding. Which question why use the lossless then, since if you end up having to transcode the lossy part, then just ripping only to the lossy bitrate will save time and space.