ABX'ed AAC 128 VBR (log posted). Angry :( Options

[Carsi]

Hi everyone.

I've been following these forums for quite a while, but today I decided to register and post my experience.

After reading on forums I decided that AAC 128 VBR would give me transparent results, so I re-ripped everything.

But today I ABX'ed Rammstein - Sonne and got 20/20. I'm so mad, I need to re-rip everything again

And I'm only using Sennheiser HD439, cheap headphones, plugged into my cheap Acer laptop. This is insane. The cymbals are what give it away every time, and there's this "swoosh" sound in general, especially in the chorus.

foo_abx 1.3.4 report
foobar2000 v1.1.18
2012/11/26 15:44:29

File A: C:\Users\Carsi\Downloads\Rammstein FLAC\Mutter\Sonne.flac
File B: C:\Users\Carsi\Music\iTunes\iTunes Media\Music\Rammstein\Mutter\Sonne 1.m4a

15:44:29 : Test started.
15:46:29 : 01/01 50.0%
15:46:49 : 02/02 25.0%
15:47:13 : 03/03 12.5%
15:48:08 : 04/04 6.3%
15:48:40 : 05/05 3.1%
15:49:30 : 06/06 1.6%
15:50:35 : 07/07 0.8%
15:52:02 : 08/08 0.4%
15:52:21 : 09/09 0.2%
15:52:50 : 10/10 0.1%
15:53:13 : 11/11 0.0%
15:53:41 : 12/12 0.0%
15:53:53 : 13/13 0.0%
15:54:15 : 14/14 0.0%
15:54:41 : 15/15 0.0%
15:55:24 : 16/16 0.0%
15:55:39 : 17/17 0.0%
15:56:16 : 18/18 0.0%
15:56:36 : 19/19 0.0%
15:56:52 : 20/20 0.0%
15:57:11 : Test finished.

----------
Total: 20/20 (0.0%)

[eahm]

Please be more specific? How did you encode to AAC? Which encoder/software?

Thanks.

[Carsi]

Please be more specific? How did you encode to AAC? Which encoder/software?

Thanks.

FLAC to ALAC in DBPoweramp and then to AAC 128 VBR in iTunes 10.7

This post has been edited by Carsi: Nov 26 2012, 16:09

[skamp]

If you still have the FLACs or ALACs, you don't need to re-rip, just mass transcode. It shouldn't take more than a few hours.

[eahm]

Perfect thank you. Do you have time to test more and would like to know where is your transparent spot? I can encode the file to different bitrates for you if you'd like me to. I'd really like to know where most people don't hear any difference between the original and the transcoded with AAC.

This post has been edited by eahm: Nov 26 2012, 16:15

[DonP]

But today I ABX'ed Rammstein - Sonne and got 20/20. I'm so mad, I need to re-rip everything again

And I'm only using Sennheiser HD439, cheap headphones, plugged into my cheap Acer laptop. This is insane. The cymbals are what give it away every time, and there's this "swoosh" sound in general, especially in the chorus.

1) If you kept the lossless, you wouldn't have to rerip.

2) Why did you do your whole collection before testing the result?

3) Things that cheap headphones don't do well are different than things that encoders don't do well. $20 can get you phones with 20+ khz response and be plenty fine. AFAIK more money gets you: better bass, smoother freq response, better isolation from external noise, comfort. AFAIK, none of that is relevant to the usual lossy artifacts.

[pdq]

Note that you may be able to hear artifacts on cheap headphones, due to their deficiencies, that you will not hear with more expensive ones (same for speakers and other hardware).

[Kohlrabi]

Why don't you just convert the one problem song/album to a higher bitrate/quality? IMHO there is no point to be obsessed about having all files at the same quality/bitrate. The main focus of lossy encoding is transparency, and you use any means to get there.

Note that you may be able to hear artifacts on cheap headphones, due to their deficiencies, that you will not hear with more expensive ones (same for speakers and other hardware).

That's still a deficiency of the codec (settings), which is (probably) adjustable by using higher settings. I don't think it's an good argument that "better" gear will mask certain artifacts. If the artifacts are there, it is not transparent with the currently used hardware, so you need to use different encoding settings to compensate for that. In the end this shows just how much lossy encoding settings are a subjective and equipment dependent choice.

This post has been edited by Kohlrabi: Nov 26 2012, 19:58

[Gainless]

I would re-rip/transcode the albums with VBR mode 5 of Winamp's AAC (~192 kbps) if you want to go safe, can't really imagine that you could still ABX it (I can't at least).

[Carsi]

I don't have the FLACs anymore, yes I'm stupid I know. Anyway, I tried a few other songs, and it wasn't as bad as "Sonne". Guess I'll just wait till iTunes Match will be released in my country then I can upgrade them all to 256 VBR. I know that may be overkill, but at least I won't notice the artifacts as in 128 VBR.

What confuses me though is that if you look at articles and listening tests on the internet in the last 2 years, almost everyone come to the conclusion that 128 AAC VBR is 99% CD quality. That certainly isn't true for me, even with no audiophile equipment....

I may have been too optimistic about 128 AAC VBR.

I'll do some more tests and post logs if I get 20/20 again like in the Rammstein song.

[BFG]

I don't have the FLACs anymore, yes I'm stupid I know. Anyway, I tried a few other songs, and it wasn't as bad as "Sonne". Guess I'll just wait till iTunes Match will be released in my country then I can upgrade them all to 256 VBR. I know that may be overkill, but at least I won't notice the artifacts as in 128 VBR.

Correct me if I'm wrong, but if you're going to go up to 256 VBR in AAC, you might as well consider other lossy options at the same time. For example -V0 LAME MP3s are around that bitrate, and transparent (or close to it).

[Nessuno]

What confuses me though is that if you look at articles and listening tests on the internet in the last 2 years, almost everyone come to the conclusion that 128 AAC VBR is 99% CD quality. That certainly isn't true for me, even with no audiophile equipment....

In fact: no one in his right mind can assure you a lossy codec is 100% transparent for everyone and every source material on earth, whatever the bitrate, and I've never seen such a claim made on HA.
There are also chances that you could spot a specific killer sample to which you are highly sensible even at the highest bitrate allowed by that codec: have you made, for example, different tests with that same track at higher bitrates?

[C.R.Helmrich]

I would re-rip/transcode the albums with VBR mode 5 of Winamp's AAC (~192 kbps) if you want to go safe, can't really imagine that you could still ABX it (I can't at least).

You could even start with VBR 4 and check how close to transparency Sonne sounds to you. Winamp's VBR-4 bitrate will probably end up near 150 kbps on that song since it's a difficult one.

Chris

[Nessuno]

By the way: how old are you? Maybe instead of re-ripping, you could just let time pass and nature make his job...

Seriously, just for the record you could keep that sample you ABXed and test from time to time in the years to come. As far as I know, there are still no long term studies about perception of compression artifacts vs aging, are there?

[kritip]

FLAC to ALAC in DBPoweramp and then to AAC 128 VBR in iTunes 10.7

What software did you rip with (not that this will cause the artifacts). I dont understand why you converted from CD, to FLAC, to ALAC, to AAC, and then remove the originals. Seems illogical to me.

[Porcus]

Hm. If I am allowed to post an ad I have no economic benefit from: Those who need big batch ripping jobs done can maybe head over to http://forum.dbpoweramp.com/showthread.php...eply-discounted .

Myself I used a Sony XL1B 200-disc changer. A couple are available on eBay now.

[Axon]

But today I ABX'ed Rammstein - Sonne and got 20/20. I'm so mad, I need to re-rip everything again

It sounds like you found a *particularly* bad problem sample here. Methinks a lot of HA regulars might want to listen to it (hopefully leading to improved encoders in the future). Could you upload a 15-30 second clip of the track to the Uploads forum?

[jensend]

If you absolutely must have perfect transparency on every single sample you will ever encode, and you will be angry at finding the slightest ABXable difference, you have two options: you can test every single song you encode to find the bitrate at which it becomes transparent for you, or you can go with lossless. Nothing else will suffice.

Imagine if everybody had access to a database containing ABX trial results from every living human listener for every sample of music ever recorded up to the present day. Would you really insist on choosing a bitrate so high it'd never been ABX'd? But even if that was so high it was no significant savings over lossless, somebody might record a piece of music tomorrow that would be ABXable at that bitrate!

I don't think that's a rational position to take. Instead, you have to realize that with any lossy encoder you face a tradeoff between bitrate and the frequency and severity of artifacts.

The only "transparency guidelines" that are useful have nothing to do with claims that the encoder is absolutely always transparent at that bitrate for every conceivable sample, listener, and test setup. Rather, a useful guideline is a bitrate at which a supermajority of normal content is no longer ABXable for a supermajority of users and at which the remaining exceptions are unlikely to have any serious adverse affects on one's listening experience.

Different people have different priorities and thus different optimal points on the bitrate vs quality curve. But for listening purposes, unless you place practically zero value on space, it's unlikely that your optimum is above 160kbps for iTunes AAC.

Given that you got rid of your originals, your disappointment may be because you were been expecting a lossy format to serve as an archival format and not just a listening format. Even if you find a setting at which all your music is transparent to you, using a lossy format for audio archival is a somewhat bad idea given the availability of cheap multi-terabyte hard drives. (1TB= ~3000 hours of stereo FLACs.) Someday there will be superior formats, and devices most likely won't play AAC forever. Transcoding from one lossy format to another is quite likely to introduce unnecessary artifacts.

[BFG]

stuff

+1 vote for adding that to the lossy section of the wiki.

[C.R.Helmrich]

Someday there will be superior formats, and devices most likely won't play AAC forever.

Both wrong.

At the high bitrates you mentioned, i.e. >= 160 kbps stereo, (xHE-)AAC and Opus are so close to the theoretical maximum in compression you can get with reasonable encoding and decoding speed that any improved codec won't sound significantly better, except on some very few signals. I promise.

Do you know any modern optical disc drives which don't play audio CDs any more? So why should devices stop supporting AAC in our lifetime? IIRC, a few years from now the patents expire, so AAC decoders can be used free-of-charge then. At least that should be true for AAC-LC, which is all you need for >= 160 kbps.

Carsi, did we overload you with answers? Have you tried Winamp's AAC encoder? I'd be curious how it does on e.g. Sonne. You can rip to FLAC or WAV with any software if the free Winamp version doesn't rip to AAC (don't remember if it does) and then use the "Send to -> Format Converter" feature.

Chris

This post has been edited by C.R.Helmrich: Nov 28 2012, 23:52

[saratoga]

The future is probably moving towards more generally re-programmable devices like smartphones, smart tvs, media centers, DAPs with apps, and so on so in a sense I think "hardware" support probably isn't very important in the long term. Dumb hardware devices that can't run user supplied media apps will certainly exist, but probably not anywhere near the extent that they used to.

[jensend]

Both wrong.

Ridiculous.

Your statements about "theoretical maximum at that bitrate" are meaningless baloney. If you had actually read my post rather than just rushing in, you'd know I'm quite aware AAC is close to transparency at >160kbps. But Opus is already a superior format which can achieve the same quality at slightly lower bitrates. It's only a small incremental improvement over AAC for stereo music (unless you need low delay), but future work will result in future improvements, at some point between 12 and 20 years from now AAC will likely look as outdated as MP2 does today.

That MP2 is reasonably close to transparent at 256kbps and up doesn't mean that it's anything but an ancient outdated curiosity given the availability of vastly better codecs that achieve the same quality at substantially lower bitrates.

I don't see mainstream portable music players with built-in support for MP2 audio anymore, and neither iOS nor Android support MP2 by default. People aren't going to leave support for long-since outdated codecs in forever. This isn't a question of patents, it's a question of the continuing effort to maintain compatibility and prevent bitrot with code that no longer sees frequent use. I'm not suggesting that this will happen to AAC any time in the near future. But if you honestly don't think it will happen in your lifetime, either you're fooling yourself or you're reaching the end of the actuarial tables.

This post has been edited by jensend: Nov 29 2012, 00:33

[eahm]

C.R.Helmrich, did iTunes upgrade to 256kbps just because people were whining? Not sarcasm, I'd just like to understand how crazy marketing is.

This post has been edited by eahm: Nov 29 2012, 00:21

[IgorC]

But Opus is already a superior format which can achieve the same quality at slightly lower bitrates. It's only a small incremental improvement over AAC for stereo music (unless you need low delay), but future work will result in future improvements

Still it need to be verificated. As for last personal test from Kamedo2 Opus is already better than AAC at 75 kbps and on par at 100 kbps. My findings are the same. Of course Opus has a lot of room for improvement.

But here I tend to agree with Chris. We probably will stick with MP3, (xHE)AAC, Vorbis and Opus for a long time.

There were a lot of effort for a new image lossy format. JPEG reigns the world.

[BFG]

There were a lot of effort for a new image lossy format. JPEG reigns the world.

And now we have PNG to complement FLAC nicely