ATRAC3plus 352kbps

[pauljones52]

I started this as I couldn't find a thread about the quality or usability of 352kb/s atrac3plus. What are your opinions on the quality compared to HI-SP and PCM. In my opinion, 352 sounds more refined in the high frequency ranges compared to HI-SP. Lack of direct CD ripping leads me to believe that it is a variant of fake SP which was experienced with Net Md's. However there would be no benefit to having a fake variant unless the 3rd gen of HI-MD's were going to play 352kb/s as a standard bitrate (I would imagine they would consider removing atrac3 48kb/s to make way for this). I read somewhere that 352 was originally engineered into the units but never released as a recordable bitrate as it was considered a poor performer when placed against other formats at high bitrates. Do you think that 352 will become a standard bitrate for future HI-MD models? (if any future ones are made). Please tell me your personal views of 352kb/s when compaired to HI-SP. Apologies if I have posted this in the wrong place as I am new to these forums

[Ishiyoshi]

Welcome to MDCF, Matt!

SQ of ATRAC3plus 352kbps has been discussed sporadically within the forum. At my end, ATRAC3plus 352kbps definitely sounds much more refined - cleaner and full of details - than ATRAC3plus 256kbps. In addition, I find ATRAC3plus 352kbps the best ATRAC3plus bit rate for use with my MDR-SA5000.

[streaml1ne]

I concur with Ishi. I also have the SA5000's and could pick up differences immediately on detailed music.

[greenmachine]

It might be slightly more detailed and artifact-less in theory, but it's not worth the extra bandwidth for me since i'm having a hard time even to abx Hi-SP from PCM with real world music/samples. Any successful abx comparisons yet or just speculation?

[pauljones52]

It might be slightly more detailed and artifact-less in theory, but it's not worth the extra bandwidth for me since i'm having a hard time even to abx Hi-SP from PCM with real world music/samples. Any successful abx comparisons yet or just speculation?

I have not been able to find any comparisons so far, however

When converting from WAV to SP in sonicstage, the resulting .OMA file stored on your computer is LP2 in size (Hence fake SP)

When converting from WAV to 352kb/s the resulting file works out at 352kb/s in size. This would indicate that 352 is a real bitrate and not just another fake SP. The general consensus that 352 is better than HI-SP, but it would be interesting to know by how much (i.e. is it worth the extra space).

[atrain]

I have not been able to find any comparisons so far, however

When converting from WAV to SP in sonicstage, the resulting .OMA file stored on your computer is LP2 in size (Hence fake SP)

When converting from WAV to 352kb/s the resulting file works out at 352kb/s in size. This would indicate that 352 is a real bitrate and not just another fake SP. The general consensus that 352 is better than HI-SP, but it would be interesting to know by how much (i.e. is it worth the extra space).

by how much is a purely subjective matter. rip a CD you know well in both bitrates & do some A->B testing. check if you can tell which is which.

with say hip-hop i'm happy with 192kbs for more complex &/or sparse [classical or say classic rock & country] i go with 256 for added clarity. for me anything more is overkill. yes i can *just* hear a slightly wider soundstage at 352 but it's not a dealbreaker for me.

Edited December 24, 2005 by atrain

[Geof]

This may be a dumb question, but where in SS do you find this 352k bitrate? I have SS 3.3 & the highest ATRAC3 bitrate I can select is 320k. Of course, the extra 32k may not be all that different, but I'd like to try it anyway. I have an MZ-RH910 if that's a factor (just got it this week.)

::EDIT:: Ah, just noticed the "lack of direct CD ripping" from the original post. If that's the reason, sorry.

Edited December 24, 2005 by Geof

[pauljones52]

::EDIT:: Ah, just noticed the "lack of direct CD ripping" from the original post. If that's the reason, sorry.

Thats why there is some confusion around the 352 bitrate, why did Sony not add direct CD ripping capabilities??

[Sparky191]

I think I can hear a difference. However HiSP is just much easier to use. In comparision to my Zen Micro or iPod Shuffle or Panasonic PCDP at 256/320 kps the HiMD at 256/352 sounds pretty much the same. If pushed I would say the Shuffle is best then Zen, then HiMD then PCDP. Not much in it though. However I suspect its as much to do with the EQ (I usually have it off) and how each device drives the earphones you're using. I'm using Mx500 ear buds which are pretty lowfi really.

[Alexx]

Personally I think ATRAC is so good compared to MP3 that old 132 is more than enough for me.

Infact for speach (eg audio cassettes), the 48kbps option works just fine.

Hi-LP is ok for personal stereo but if i was going to link it to my car stereo its better @ 132

Can I just ask if anyone can feel a diffrance between good old 292 and 352

[Sparky191]

Personally I think ATRAC is so good compared to MP3 that old 132....

Thats meaningless unless you know what bitrate, encoder and encoder settings you are using. That makes a world of a difference.

[Alexx]

Personally I think ATRAC is so good compared to MP3 that old 132 is more than enough for me.

Infact for speach (eg audio cassettes), the 48kbps option works just fine.

Hi-LP is ok for personal stereo but if i was going to link it to my car stereo its better @ 132

Can I just ask if anyone can feel a diffrance between good old 292 and 352

sorry what are those things?

I am reffering to taking a CD and putting it to MD with sonic stage.

hope that makes sense

[Sparky191]

What things?

[Syrius]

I think he's referring to cassettes.

[Sparky191]

I think he's referring to cassettes.

ah...doh!

[pauljones52]

Thanks for everyone who voted on this, it seems that most people can tell the difference between the two. I've noticed that 352 sounds alot better with music with lots of treble (e.g. dance music). But whether its worth the extra space is another matter. I personally dont like too many albums on one disc as I forget whats on there, so I have used 352 alot.

Matt

[Sparky191]

You need another poll now for SP vs 352kps.

[greenmachine]

...it seems that most people can tell the difference between the two.

I'd rather say it seems that most people think they can tell the difference between the two.

Can they without looking at the bitrate?

Edited January 11, 2006 by greenmachine

[Sparky191]

In fainess some people think 64kps is ok

[pauljones52]

You need another poll now for SP vs 352kps.

I'll leave that one to you sparky as I do not have any units that can record SP anymore (the record function broke). Hopefully sonicstage 3.4 can transfer true SP, then I would be able to test.

[Rumz]

After hearing some hype about 352kbps I gave it a whirl (not on Hi-MD) and I have to admit that for me it has the edge that SP had. Hi-SP @ 256k is pretty good, more than enough for most music, but for certain types (like my belovid film scores ) the 352kbps bitrate is absolutely heavenly.

Whether you can tell the difference is one thing, how much difference you can hear might be another still. And yet another issue is whether those differences are worth it to you to have larger files. To each his own-- I'll be re-ripping my soundtracks at 352kbps eventually. This is why I still use Atrac.... gapless and sounds amazing.

[The Low Volta]

anyone tried an ABX-test and achieved really statistically significant results saying they can tell the difference between 256 and 352? I know I can't for most types of music... beware of placebo-effects

[Sparky191]

anyone tried an ABX-test and achieved really statistically significant results saying they can tell the difference between 256 and 352? I know I can't for most types of music... beware of placebo-effects

Any tests I tired, I could but only on specific CD's that are qood quality and certain types of music. Like an unplugged session or classic stuff. Some old rock stuff I couldn't tell a difference.

[greenmachine]

If you listen very hard with good equipment in a quiet environment, maybe you can tell the difference between the three [A292 (SP), A3+256 (Hi-SP), A3+352]. Most of the time i don't listen that hard though and as a portable format (in noisy environments with portable headphones) all of them seem to be (more than) adequate. It's - as usual - just a matter of where you set your threshold. The music itself is by the way still more important than the codec for me. I'd rather listen to a good song in Hi-LP than to one i dislike in PCM.

[Sparky191]

Why would you listen to music you don't like?

Personally the fact I like a song more, would make a low bitrate even more irritating.

[greenmachine]

I wouldn't if i didn't have to, but if i could get a copy of a certain song/album in low quality only, i'd be happy to get it at all. I don't primarily listen to music because of it's fidelity but rather due to its content. Fidelity comes second.

[Sparky191]

Yes but you said listen to music you don't like....

[ZosoIV]

Just tried the 352kbps setting with several "difficult" John Coltrane tracks over Grado SR-325 headphones and an M-Audio 96kHz soundcard, and the difference between it and Hi-SP (256kbps) is definitely hearable (and reproducible) on certain passages, especially when John blows into his sax really loud and at high pitches. There's less ringing/smearing in the 352kbps file; for example, the sax has loses detail in some passages with Hi-SP, but not with 352kbps. Also, there's much less, if almost no pre-echo (smearing of very sharp cymbal hits) in the 352kbps version, while I can occasionally discern some with Hi-SP. (For fun, try to encode a "castanet" sample with Hi-SP!!). The 352kbps also retains the original hiss/air of the PCM file, whereas the Hi-SP file loses some. Thus, I've concluded that 352kbps is the setting to use for "transparent" encodings that sound just like the PCM original. Too bad it takes such a high bitrate for this with A3+…..for example, LAME, Ogg, and MPC reach this point around 200kbps or less.

Keep in mind that because of time constraints, I did not do a double-blind ABX test here, so YMMV - I only chose to listen to parts of songs that traditionally give ATRAC lots of trouble. For most passages, I doubt there would be a hearable difference.

Oh, forgot to mention that I also compared some 352kbps encodings with some ATRAC Type-R ones I made with my MDS-JE530 deck (over the S/PDIF connection). For example, Type-R makes the acoustic guitar strums on the end solo section of Pink Floyd's "Comfortably Numb" sound thicker and less certain (a pre-echo issue). A3+ sounds more stable here and is crisper than Type-R, it would seem, but otherwise, they are very close in terms of sound/coloration. Type-R always had a certain "punchiness" to it that Hi-SP doesn't seem to have, but the higher 352kbps setting does.

I think I've finally found a replacement for SP

[Sparky191]

I thought it similarly better aswell. Any thoughts about 352kps vs MP3 320kps Lame High Quality? Or even 256kps ATRAC vs 320Kps Lame HQ.

[ZosoIV]

I thought it similarly better aswell. Any thoughts about 352kps vs MP3 320kps Lame High Quality? Or even 256kps ATRAC vs 320Kps Lame HQ.

LAME is an extremely well-tuned codec: despite MP3's format limitations (poor time resolution, crude joint-stereo switching, >16kHz coding issues, etc), it still is competitive with newer formats like AAC and Ogg at bitrates above 128kbps. Considering how poorly ATRAC3/ATRAC3plus has performed in the few blind tests I've seen on the internet (and the couple I've done myself by merely having somebody else switch sources back and forth on my deck/DAC), LAME is probably better at 256 or 320kbps by simple virtue that it has been constantly tuned and tweaked for the past 5 years. ATRAC3plus doesn't get that sort of attention because it's closed source, so although it may be a better format on paper, it may never show its full potential. LAME, on the other hand, represents MP3 at the extremes of what it is capable of; I doubt they can get it sounding much better than where it is today.

[pauljones52]

I second ZosoIV's comments. I would also add that lame has a great advantage in that it has a variable bit rate. I only wish that the mp3 playback in the second gen minidisc units was not flawed.

I feel that the people who developed lame to such a high standard are legends and I want to give them money just for keeping it free!

Edited January 25, 2006 by Matt J

[Sparky191]

I like ATRAC on MD I think it sounds great. In my own blind listening tests between high bitrates ATRACs or LAME MP3's I'm not hearing any significant differences between the encodings, in terms of errors or artifacts. I do notice more of a difference of voicing, and colouring to the sound. But I think this is mainly due to physical device rather than encoding (ATRAC/MP3) I use.

So simply for ease of use I've tended to stick to high bitrate MP3's CBR because they the most compatible across all the devices I've owned. If I want to use my HiMD as a player, I'll rip the track from the CD in ATRAC. I don't do transcoding. If my HiMD was my only player I might do it differently.

I do like the 352kps though. I definately use it in preferance over HiSP for my favorite CD's and Acoustic and Classical Albums.

[ZosoIV]

....But I think this is mainly due to physical device rather than encoding (ATRAC/MP3) I use.

I do like the 352kps though. I definately use it in preferance over HiSP for my favorite CD's and Acoustic and Classical Albums.

That's a good point - MD players do have a certain sound signature, especially those with so-called HD amps, so that might account for some of the coloration differences.

I'd say that for 99% of music, MP3 and ATRAC3plus are probably indistinguishable at high bitrates. ATRAC might cause errors on different signals than MP3 (or vice versa), but more likely, signals that cause one to trip will also cause any audio codec trouble. Examples of this are electronic/vocoder sounds, bells tolling, castanets, harpsichords, and trumpets, all of which are very hard to code with lossy compression because their signal components are so complex and random.

I will probably only use 352kbps on things I will listen to while lying down or relaxing - everything else will be just fine with Hi-SP. As a side note, do the hardware units encode Hi-LP/LP2/Hi-SP any better than SonicStage? I am getting my MZ-NH1 on Friday.

[sfbp]

The problem for myself (and therefore most people I would guess) is that the only way to get to 352 kbps is by ripping a CD to PCM, and then transcoding it to 352kbps.

However as I have maintained for a while now, using Sony's high-speed ripping, you get junk. Therefore I would not expect the 352 kbps to be the least different from many worse bitrates that all sound dreadful after transcoding from ripped PCM (to each one).

(disclaimer: I even like some, maybe even most, music in LP4, depending on how it is prepared so you can discount my comments entirely on that basis).

The one thing I WOULD expect to work as a valid way of getting 352kbps is to rip to Atrac Advanced Lossless (AAL) with the import setting to 352kbps. If the result of transferring to HiMD starting from that file sounds different, then I will stand corrected.

But sound imported/compressed as AAL from any legitimate (WAV, properly ripped CD) source and then transferred to hiMD at 256k (Hi-SP) sounds better (to my ears) than imported PCM (assuming PCM is transcoded subsequently down to 256k). So you need to compare apples to apples and oranges to oranges.