HydrogenAudio

Lossy Audio Compression => Other Lossy Codecs => Topic started by: birdie on 2010-08-14 20:29:37

Title: Sony's ATRAC fate
Post by: birdie on 2010-08-14 20:29:37
Has anyone ever compared ATRAC3+ to HE-AAC2 at, say, 128Kbit/sec?

And the second question is, why Sony almost abandoned this codec even though it's at least superior to MP3 and Ogg Vorbis? Yes, Sony promotes using this codec in its products, but ATRAC has never been submitted to any standards body, thus it's just a closed codec with zero spread.

I remember when I coded 9 years ago some songs for my then digital Sony player in then probably ATRAC2 96Kbit/sec format, it sounded as good as MP3 at 192KBit/sec.
Title: Sony's ATRAC fate
Post by: greynol on 2010-08-14 20:39:19
Either provide ABX results or don't make comparative claims about quality.
Title: Sony's ATRAC fate
Post by: lvqcl on 2010-08-14 20:43:21
And why HE-AACv2 instead of plain LC-AAC? HE-AACv2 was designed for very low bitrates.
Title: Sony's ATRAC fate
Post by: birdie on 2010-08-14 20:50:20
I'm not an expert in AAC, so I thought "High Efficiency" must be better than "Low Complexity"
Title: Sony's ATRAC fate
Post by: dv1989 on 2010-08-14 20:59:06
It may have been an idea to check that first.

HE-AAC is designed for low bitrates. Its 'efficiency' is really just it reconstructing high frequencies from lower ones it stores.
Title: Sony's ATRAC fate
Post by: saratoga on 2010-08-14 21:27:30
And the second question is, why Sony almost abandoned this codec even though it's at least superior to MP3 and Ogg Vorbis?


Because its an obsolete hybrid subband/mdct codec more like MP3 then anything else:

http://wiki.multimedia.cx/index.php?title=ATRAC3plus (http://wiki.multimedia.cx/index.php?title=ATRAC3plus)

MP3 survives because it became the standard for compressed audio and everyone supports it, so it made sense to keep using it even if pure MDCT codecs can compress more efficiently.  ATRAC never caught on, so theres little sense promoting it when you can just use more widely supported and technically superior modern formats and drop all the baggage.

I'm not an expert in AAC, so I thought "High Efficiency" must be better than "Low Complexity"


Its lower quality but higher compression.  Efficiency refers to how well it compresses. If you don't need maximum compression, 'low complexity' gives higher quality.  Its call 'low complexity' because 'high efficiency' and the now forgotten 'main' AAC variants are/were much more computationally intensive.
Title: Sony's ATRAC fate
Post by: Alex B on 2010-08-14 22:24:41
Has anyone ever compared ATRAC3+ to HE-AAC2 at, say, 128Kbit/sec?

ATRAC3 132 kbps (not 3+) was tested in Roberto's public multiformat 128 kbps test a few years ago:
Presentation: http://listeningtests.t35.com/html/Multifo...resentation.htm (http://listeningtests.t35.com/html/Multiformat_128kbps_public_listening_test_presentation.htm)
Results: http://listeningtests.t35.com/html/Multifo...est_results.htm (http://listeningtests.t35.com/html/Multiformat_128kbps_public_listening_test_results.htm)

ATRAC3 was the clear loser. The other tested formats were iTunes AAC, LAME MP3, Ogg Vorbis aoTuV, Musepack and WMA. Vorbis and Musepack were tied at first place.

If I have understood correctly the newer ATRAC3+ was originally intended for lower bitrates like 64 kbps. However, the recent Sony encoders (e.g in Sound Forge) can encode ATRAC3+ at high bitrates up to 352 kbps stereo and 512 kbps 5.1. I have no experience of its quality.
Title: Sony's ATRAC fate
Post by: kornchild2002 on 2010-08-14 22:55:19
Additionally, Sony had abandoned the format here in the U.S.  You have to go through a series of specific steps and actually active ATRAC playback on Sony's own PS3 and their current line of Walkman players don't even support the format (some of them do in Japan though).  I don't know why anyone  even worry about testing ATRAC3+ in an actual listening test when you can't even buy a portable player, here in the U.S., that supports the format.  I don't think anyone should even be contemplating ATRAC these days especially since the developing company has stopped supporting it.  Gone are the days when you had to use the format to copy media to a Sony labeled portable player.
Title: Sony's ATRAC fate
Post by: saratoga on 2010-08-14 23:12:52
Has anyone ever compared ATRAC3+ to HE-AAC2 at, say, 128Kbit/sec?

ATRAC3 132 kbps (not 3+) was tested in Roberto's public multiformat 128 kbps test a few years ago:
Presentation: http://listeningtests.t35.com/html/Multifo...resentation.htm (http://listeningtests.t35.com/html/Multiformat_128kbps_public_listening_test_presentation.htm)
Results: http://listeningtests.t35.com/html/Multifo...est_results.htm (http://listeningtests.t35.com/html/Multiformat_128kbps_public_listening_test_results.htm)

ATRAC3 was the clear loser. The other tested formats were iTunes AAC, LAME MP3, Ogg Vorbis aoTuV, Musepack and WMA. Vorbis and Musepack were tied at first place.


ATRAC3 is quite different then ATRAC3+.  They're both hybrid subband codecs with a lot of weird features that never made it into modern formats (gain control, extraction of tonal and spectral waveform components for quantization, etc).  But ATRAC3+ appears to change quite a lot.  The subband decomposition is changed from working vaguely like how SBR works in AAC-HE to being more like how its done in MP3 with a lot of small bands instead of a few big ones.  Theres better stereo coding, and huffman/vector coding is much more sophisticated.
Title: Sony's ATRAC fate
Post by: birdie on 2010-08-15 07:21:02
ATRAC3 132 kbps (not 3+) was tested in Roberto's public multiformat 128 kbps test a few years ago:
Presentation: http://listeningtests.t35.com/html/Multifo...resentation.htm (http://listeningtests.t35.com/html/Multiformat_128kbps_public_listening_test_presentation.htm)
Results: http://listeningtests.t35.com/html/Multifo...est_results.htm (http://listeningtests.t35.com/html/Multiformat_128kbps_public_listening_test_results.htm)


Thanks for these results, I have no questions about ATRAC any longer.