QAAC: discussion, questions, feature requests, etc., [originally a thread for a feature request] Options

[eahm]

Don't care about bitrates. 160 kbps silence is not "better" than 32 kbps silence. There is easily compressible sound which you can't directly compare with harder "noise". If the quality mode ensures a threshold of maximum loss, trust in it.

Where is the "Like" button?

This post has been edited by eahm: Jun 9 2012, 20:13

[IgorC]

Constrained VBR (CVBR) might be a good solution. It doesn't go as low as TVBR does occasionally and causes some artifacts as here http://www.hydrogenaudio.org/forums/index....st&p=768156

LigH,
It's strange. Don't You watch the football righ now?

This post has been edited by IgorC: Jun 9 2012, 20:41

[LigH]

I do. And I am not really satisfied... But that's quite usual.

[eahm]

[qaac] release 1.39 (refalac 0.50)
posted 6 hours ago by nu 774

- Support "REM DISCNUMBER" "REM TOTALDISCS" in cuesheet.
- Flush stdio buffer when stdout is connected to a pipe.
- Update mp4v2 to svn rev 496.

https://sites.google.com/site/qaacpage/cabinet

[Meeko]

Took me a while to get this to work, but once I got it working in foobar, its great. The new change to no longer need Quicktime helped because I could never get quicktime to work properly on my machine. Thanks for all your hard work nu774.

[eahm]

So, I was trying to transcode with qaac on an older laptop and it didn't work and I don't know why precisely.

The laptop has Windows 7 32bit, both iTunes and QuickTime latest version are installed, I've tried to copy CoreAudioToolbox.dll and CoreFoundation.dll on the same folder where qaac.exe is but nothing. The software I am using to convert with qaac is foobar2000 1.1.14 beta 1 and the error message I receive is: "Conversion failed: The encoder has terminated prematurely with code -1073741515 (0xC0000135); please re-check parameters". I have .NET 4.0 installed, do I need an older version as well? I've not yet tested qtaacenc.

Thanks.

This post has been edited by eahm: Jul 29 2012, 21:26

[lvqcl]

I've tried to copy CoreAudioToolbox.dll and CoreFoundation.dll on the same folder where qaac.exe is but nothing.

Remove them, and copy msvcp100.dll and msvcr100.dll to the qaac folder.

[eahm]

I've tried to copy CoreAudioToolbox.dll and CoreFoundation.dll on the same folder where qaac.exe is but nothing.

Remove them, and copy msvcp100.dll and msvcr100.dll to the qaac folder.

It worked thanks. Shouldn't Visual C++ be added to the list of requirement on the qaac homepage?

[LigH]

It may have been part of earlier packages...

[lvqcl]

qaac 1.39 package contains: qaac.exe, refalac.exe, libsoxrate.dll, msvcp100.dll, msvcr100.dll.

[Anakunda]

Hello, why qaac can't use HE with TVBR, and is therefore less suitable for VBR mode at ~64kbps than NeroAAC?
Probably --cvbr 64 --he will produce something..but is the quality at least as good as Nero's -q 0.25 ?

[nu774]

Hello, why qaac can't use HE with TVBR

Because HE-AAC encoder of Apple CoreAudio/QuickTime doesn't have TVBR mode.

[IgorC]

Hello, why qaac can't use HE with TVBR, and is therefore less suitable for VBR mode at ~64kbps than NeroAAC?
Probably --cvbr 64 --he will produce something..but is the quality at least as good as Nero's -q 0.25 ?

What makes You think that Nero's VBR is any better than Apple CVBR? And why do You think CVBR is necessary worse than TVBR?

[Anakunda]

Hello, why qaac can't use HE with TVBR, and is therefore less suitable for VBR mode at ~64kbps than NeroAAC?
Probably --cvbr 64 --he will produce something..but is the quality at least as good as Nero's -q 0.25 ?

What makes You think that Nero's VBR is any better than Apple CVBR? And why do You think CVBR is necessary worse than TVBR?

Thanks. Is CVBR HE at 64kb better than Nero VBR HE at ~64kb? I've been using Nero q 0.25 long time but it might increase the bitrate when necessary while CVBR always keep the given bitrate which may lead to worse quality on complex music?

[IgorC]

I've been using Nero q 0.25 long time but it might increase the bitrate when necessary while CVBR always keep the given bitrate which may lead to worse quality on complex music?

CVBR isn't CBR, and it does increase bitrate where it's necesarry.

For You the variation of bitrate is the most important and only indicator of quality?

[Anakunda]

I'm not sure Does it mean that QuickTime's CVBR at 64k is always better than Nero at q 0.25?

[LigH]

This is entirely a subjective questions. Listen to hundreds of samples and decide yourself...

[IgorC]

This is entirely a subjective questions. Listen to hundreds of samples and decide yourself...

Or google for the latest public test (2011) of Apple and Nero AAC encoders at 64 kbps.

It could be understandable to mention Nero if it was at least an average AAC encoder. But Nero was the last and worse than Coding Technologies, FhG/Winamp and Apple encoders in last AAC public test.
Furthermore Nero is outdated. The last fixes were made in 2009. The last quality improvements are dated by 2007.
It's gone. Understand it.

This post has been edited by IgorC: Aug 1 2012, 14:00

[neothe0ne]

1) the TVBR value doesn't correspond to the encoder settings. ie. -V 90 (or default) registers as q91, -V 75 as q73, etc.

At the interface level, TVBR quality parameter accepts values from 0 to 127.
However, the QuickTime AAC encoder actually has only 15 quality steps. Therefore, parameter is get rounded to the nearest functional value, which is saved into the "tool" tag.

Can you share what those 15 functional values are? I could test myself (since I obviously won't be using all 15 ever) but if you have the information handy it would save me some time

[LigH]

Here is a graph by 'kamedo2':


qaac tvbr number-bitrate relations (Hatena Fotolife gallery)

I know I saw even a better one, but don't remember where. Probably on the qaac site or even here; someone explained the change in the quality steps between specific generations of the QuickTime AAC codec.
__

IgorC posted it in a thread about qtaacenc.

This post has been edited by LigH: Aug 14 2012, 07:44

[Anakunda]

I havenot tried all presets but they seem to be step 9:
1, 10, 19, 28, 37, 46, 55, 64, 73, 82, 91, 100, 109, 118, 127

Btw. is that true that tvbr 82 corresponds to Vorbis's q 5.0,
tvbr 91 corresponds to Vorbis's q 6.0, tvbr 100 corresponds to Vorbis's 7.0 etc?
About bitrate they're approx comparative:

tvbr 91 ~192kbps
tvbr 100 ~224kbps
tvbr 109 ~272kbps
tvbr 118 ~ 320kbps
tvbr 127 ~352kbps

[LigH]

Btw. is that true that tvbr 82 corresponds to Vorbis's q 5.0,
tvbr 91 corresponds to Vorbis's q 6.0, tvbr 100 corresponds to Vorbis's 7.0 etc?

In general ... rather "no", because Vorbis is a quite different algorithm, it does not work in the same way as AAC, and I doubt that Apple made efforts to synchronize their quality levels to any other software.

But subjectively you will probably be close in quality and bitrate. And remember, you can fine-tune Vorbis with fractional quality values.

This post has been edited by LigH: Aug 14 2012, 08:41

[neothe0ne]

I've encoded about a thousand tracks and haven't batted an eye at TVBR (got lots of stuff from 170 to 200 with V91), but I just encoded PSY's 6th album (Korean electronic/dance pop) and WOW.
Just looking at the title track, Gangnam Style,
tvbr 91 ~171 kbps
iTunes Plus ~260 kbps
tvbr 109 ~233 kbps

I realize that -91 doesn't guarantee anything near 192, nor -109 around 270+, but then I compared the bitrate distribution:
CVBR 256 starts at ~140 kbps and bilds to ~200 kbps, and at 6 seconds in, hits ~450 kbps. (6 seconds in it changes from low/bass beat to add a vocal)
tvbr 91 starts at ~100 and is constant until it hits ~130-160 at 6 seconds in.
tvbr 109 starts ~130 and is constant until it hits ~170 at 6 seconds in.
(studied in foobar2000 with 3 VBR-updates per second)

My first thought was tvbr had a lower frequency cutoff than iTunes Plus, but turns out that isn't correct. (tvbr 91 does have a soft wall at ~19.4 khz, whereas the other two go up to 22 khz, but the first 6 seconds in question don't have any (visible) frequencies going high enough to have been cut off anyway)

CVBR has a higher floor than TVBR, but anyone have any idea what accounts for the nearly 300 kbps difference in the ceiling, which I hadn't believed would differ significantly between CVBR and TVBR?

[Anakunda]

For curiosity I've made spectral graphs of TVBR versus Vorbis encode both at almost same bitrate 192k and from these the Vorbis looks like much better as maintaining higher cutoff range. Anybody can confirm that the Vorbis encode keeps more of the original?


TVBR 91 (full / zoomed / frequency):


Vorbis q6 (full / zoomed / frequency):

[jetpower]

For curiosity I've made spectral graphs of TVBR versus Vorbis encode both at almost same bitrate 192k and from these the Vorbis looks like much better as maintaining higher cutoff range. Anybody can confirm that the Vorbis encode keeps more of the original?


TVBR 91 (full / zoomed / frequency):


Vorbis q6 (full / zoomed / frequency):

AAC has less "holes" in high freq, also are you a bat?

This post has been edited by jetpower: Aug 14 2012, 21:18