New Listening Test in 2007, MP3 @ 128 kbps or Multiformat @ 64 kbps? Options

[eofor]

Well, I suppose that way we can defeat the claims about "CD quality at half the bitrate of 128kbps MP3!"
I'd still like a 96kbps test though.

I think that claim will be harder to defeat than you think. Even at 48k some of the samples were near transparency with HE-AAC, and I doubt that WMA Pro 64k is worse than HE-AAC 48k. Microsoft never claimed that 64k WMA is "CD quality" for every possible sample, remember?

[le_canz]

Voted for 64 kbps

I am curious about WMA10 pro and HE-AAC.

[zombiewerewolf]

I voted for 64kbps. Though, bitrate is close to the last test, I think it much more useful than 80/96kbps. Most of online video or music usually use 48, 64 or 128kbps. I rarely see ones broadcast with 80/96kbps.

My second preference would be mutiformat @128kbps but the test would be very difficult to do.

[Sebastian Mares]

Well, the last multiformat test at 128 kbps is pretty recent.

[robert]

I voted 64k test, because I would like to have 3.98 stable ready before a 128k test comes.

[Firon]

I think that claim will be harder to defeat than you think. Even at 48k some of the samples were near transparency with HE-AAC, and I doubt that WMA Pro 64k is worse than HE-AAC 48k. Microsoft never claimed that 64k WMA is "CD quality" for every possible sample, remember?

My ears aren't quite as good as guruboolez' (guruboolez's?), but there were only two samples where I ranked at least one of the 48k contenders above 3~3.5 (and it still wasn't a 5), which is certainly not transparent. The vast majority of the samples were awful.

64k could be better, but I don't really think it will be -significantly- better than 48.

This post has been edited by Firon: Dec 18 2006, 02:50

[vinnie97]

In for 64.

[Remedial Sound]

My vote is for a 96 kbps test. While once a benchmark for "acceptable" quality, modern encoders have evolved to the point where 128 kbps is transparent to most listeners for most music. I'd be in interested in seeing whether the great work of developers has progressed audio compression to the point where transparency can be achieved under 100kbps (and which codecs are best at achieving this). This would also be of particular interest to owners of flash-based portables looking to have [close-to-] transparency at the lowest possible bitrate.

[Trondis]

Hi!

I vote for 64kbs, but don't leave out Atrac3Plus this time! I use that on my minidisk, and I want to know how it comes out compared to others.



Trondis

[Sebastian Mares]

Minidiscs don't support ATRAC3+ as far as I know.

[singaiya]

Even though 64kbps seems close to 48, it's good to remember that it's a 33% increase. Those 16 bits probably go a lot farther (in terms of perception) than adding 16 bits to say 128. (That's speculation, but seems reasonable to me.)

[Kef]

Judging by the quality increase going from 48 to 64 kbps using ogg vorbis (aotuv b5) I'd say a 64 kbit test would be very interesting. As someone else pointed out it is a 33% increase of the bitrate and I would be curious to find out which codec would benefit the most of it. Since we just had a 48kbps test using the same contenders would make sense.

[vitos]

I would vote for 80 or 96 kbps multiformat test. Considering great results of current encoders at 128kbps, I am interested in their performance in 25% lower bitrate range. From my observations 64kbps still means "very good quality internet radio" for me, but that's not enough for portable usage and pleasant listening on headphones. I expect the results would be too much correlated to those from 48kbps listening test.

This post has been edited by vitos: Dec 22 2006, 16:41

[Caroliano]

I voted for 64kbps, and I still would be interested in test that, mainly because vorbis vs HE-AAC. M$ claim wold also be interesting.

Nevertheless, I prefer an 80kbps one, because we can see the level of transparency in this bitrate (something not adequate for 64kbps), and test some Mp3 codecs, as lame. The 80kbps one would be very good for evaluate codecs for portable use. The downside is that we would need an pre-test for LC vs HE AAC.

[muaddib]

For 80 kbps LC AAC there is already "pretest" comparing PNS and IS in Nero codec.

[Junon]

The downside is that we would need an pre-test for LC vs HE AAC.

I'd simply include them both into the test. This way the user could see whether it made sense encoding to HE-AAC in the 80 kbps bitrate range. Especially portable players would benefit of that, since no or only very slight differences in quality clearly caused the less decoder-intensive LC-AAC to be way more desirable. For the Nero developers this would also be a useful indicator about the question if their encoder should default to HE-AAC or LC-AAC in case someone used the -q 0.3 commandline setting.


Edit:

For 80 kbps LC AAC there is already "pretest" comparing PNS and IS in Nero codec.

Yes, but this test won't answer the question whether LC-AAC or HE-AAC is the better choice for a 80 kbps file.

This post has been edited by Junon: Dec 22 2006, 17:06

[muaddib]

For 80 kbps LC AAC there is already "pretest" comparing PNS and IS in Nero codec.

Yes, but this test won't answer the question whether LC-AAC or HE-AAC is the better choice for a 80 kbps file.

I guess that question if PNS or IS give any advantage when used in LC should first be answered. Then testing LC against HE can be done. But it seems that question LC against HE at 80 kbps will be answered only after this 64 kbps test

[Junon]

I guess that question if PNS or IS give any advantage when used in LC should first be answered. Then testing LC against HE can be done. But it seems that question LC against HE at 80 kbps will be answered only after this 64 kbps test

That, of course, is true. Too bad the poll doesn't include any choice about the bitrate range to be used in case someone voted for a multiformat test instead of the MP3 one.

This post has been edited by Junon: Dec 22 2006, 17:16

[singaiya]

I don't vote for a 96 kbps test because in the 48 kbps test I could only identify the high anchor twice. It got a 4.5 and a 4.8 from me, and in the overall group results it received 4.7. I think most contenders at 96 kbps would be so close to the top, that statistically confident ranking of x being better than y is highly unlikely.

[The Sheep of DEA...]

64kbps for me! aoTuV b5 with raised lowpass is a whole different ballgame than 48kbps Vorbis. 64kbps HE-AAC shouldn't surprise anybody (some folks claim they can pick out the SBR any day). WMA10Pro at 64kbps would be an interesting candidate, especially if it can be tested in VBR mode.

As for VBR WMA(pro), why can't one person (the test administrator) make all the encodes of full songs in VBR, cut them down to sample-size, and distribute all the [lossless re-encoded] pieces for testing? That would make a lot more sense than to force an encoder to do CBR mode when it is capable of a much more advanced mode of bit allocation.

This post has been edited by The Sheep of DEATH: Jan 12 2007, 23:59

[Sebastian Mares]

Maybe because the test administrator doesn't have enough money to buy 20 CDs (it has to be CD or since I don't know of any online store that offers lossless downloads) or doesn't want to get into trouble by illegally downloading the tracks.

This post has been edited by Sebastian Mares: Jan 13 2007, 08:56

[herefornow]

Would like to see 64kbs test. Tried the 80kbs AAC recent test and am amazed at what can be done with so little. My portable only supports mp3 and wma. Here's to progress!

cheers,
herefornow

This post has been edited by herefornow: Jan 13 2007, 09:48

[kornchild2002]

I voted for 128kbps. My player doesn't support HE-AAC or WMA pro so I would like to see a multi-format test conducted again at 128kbps (or at 96kbps).

I receive about one iTunes gift card every odd week so it would be nice to know how that (128kbps AAC) is stacking up against other encoders at the same setting. Correct me if I am wrong but, wasn't the last 128kbps multi-format test conducted over a year ago (around the holiday season of 2005)?

[halb27]

What about a mixed test: ~128kbps mp3/mpc and ~96kbps aac/vorbis/wma? Should be a roughly competetive constellation.
Sure a big test as a series of mp3 encoders is to participate. But also one of major practical interest IMO which should be allowed some time to proceed.

[/mnt]

A new Mp3 test could show if or how much iTunes + Fhg + Xing (Helix) mp3 encoders have improved since most average computer users (n00bs) still use software that uses those encoders.

This post has been edited by /mnt: Jan 13 2007, 15:34