Dear friends,

I downloaded QT6.0 today and fortunately, my old reg. key for
QT5.0 upgraded it to pro version. So, I decided to test its MPEG-4
Audio encoder implementation. It seems that they have restricted
it to AAC Low Complexity codec. I started my listetning test
with my Grado SR-80 headphone. I hadn't got enough time to do really
careful listening tests and I had to restrict my tests to 128 kbps bitrate.
Anyways, my test cases and the results are as follows:

Test Case: Eternal Tears Of Sorrow - Autumn's Grief (Metal)
Settings: QT6.0 AAC-LC 128 Kbps
Results: Descent Frequency Response (No lowpass filter; altough the
frequnecy response from 16 khz to 22 khz has less intensity) Less detectable
pre-echo compared to the other competeing codecs (Psytell AAC, OGG Vorbis 1.0,
LAME 128 CBR alt-preset), Slight detectable flange, Perfect stereo image.
compared with the other codecs at this setting, QT6.0 AAC's overall quality is
better IMO.

Test Case: Castanets
Settings: QT6.0 AAC-LC 128 Kbps
Results: Almost the same amount of pre-echo compared to Psytell AAC,
MusePack and OGG Vorbis 1.0; Psytell AAC and MPC produce an annoying
flage-like effect in this case. OGG is probably the winner here.

Test Case: Spahm
Settings: QT6.0 AAC-LC 128 Kbps
Results: Lots of noise like other codecs. Each codec has its own
problem; but the overall quality of QT6.0 is obviously better compared
with the other codecs (128 kbps!)

My Conclusion: The overal quality (esp the first test case) has really
impressed me. I give this codec an excelent mark for 128 kbps
streaming purposes. Please let me know if you've done any other tests on
this codec. AAC has a really good potential for further developments for sure!

Cheers,
Mehrtash

Well, it has been rumored that QuickTime 6 uses the FhG professional codec (AACdemo2.2) with some weird modifications implemented by Dolby to increase speed with small quality sacrifice.

So, no wonder it sounds so good.

Regards;

Roberto.

If you recently purchased a reg key for version 5.0 it would unlock the version 6.0 preview, but a new key must be purchased if you want to unlock the released version!

I know this from first hand fucking experience.

ff123

I have downloaded the 6.0 final version from www.apple.com/quicktime
It mustn't be a preview edition... is it?

PS> I have two keys around here... One of them is only for 5.0 Pro
(once I entered it, It said it is for 5.0 and doesn't work for 6.0) but the
other one works for future versions too... So did with my 6.0!

Cheers,
Mehrtash

I doubt about that... FhG guys aren't that lazy to let their refrenece encoder
stay the same for years... Anyways, the codec sounds far better than
Liquid Audio 4.0 (5.0) AAC implementation at 128 kbps. There must be
some quality-wise tweakings done in this revision Maybe due to
Dolby modifications (I doubt!) or maybe due to FhG tunings...
I haven't got any AACdemo 2.2 sample in my hands to compare its artifacts
with QT6.0's, But I'm sure QT6.0's much more optimized than Liquid's.
(Is Liquid Audio 4.0 codec the same as AACdemo 2.2's?)

Let me know if you've got a real clue about it... I don't like rumours!

PS> Please don't get me wrong; I didn't implied it sounds so good;
If fact, I meant to say the overall mark is perfect among other available
codecs at 128 kbps.

Edit: Added the PS!

Cheers,
Mehrtash

Yes, they are.

Indeed, they stopped working on MPEG4 audio quite some time ago, and now they are mostly working on MPEG7.

AACdemo has been frozen for years. The same applies to their MP3 reference encoder, MP3enc 3.1



No.
Liquid Audio uses FhG AACdemo 3.0, it's their consumer encoder.
QuickTime (seems to) use AACdemo 2.2, the professional encoder.

It's like Psytel AACenc vs. Fastenc, or FhG MP3enc 3.1 and Fastencc. Professional vs. Commercial. Slow vs. Fast.

Regards;

Roberto.

How do you people get to the AAC options? I regged QT6 now and I still can't save AAC content in Soundforge. Actually all codecs except for AAC show up.

Try creating a few clips using the application pc_drummer. I think there is a free evaluation copy which you can download from the internet. The clips created are normally very tough on the transient detection and compensation.


KWW

QuickTime codecs defer from Windows ACM codecs; You have to load your
content into your QT6.0 Pro and export it into your prefered format; (QuickTime
Movie -> Audio -> MPEG4-Audio AAC-LC)

Cheers,
Mehrtash

Heh, I'm not dumb!

That's what I tried, and it just doesn't show AAC.

You said you tried SoundForge but what I said was for QT Player ???
I do not use SoundForge anyways and there's no QT exporting support
in CoolEdit Pro 2.0. I guess SoundForge uses its own (restricted)
QT exporting plugin which must be updated too or maybe apple
has limited the usage of AAC inside QT Player.

Cheers,
Mehrtash

The Player allows transcoding?? Ok, gonna try. Thx!

--edit--
Let's see if the .MP4 with AAC audio are conform.

It sounds OK so far, gotta make maybe some blind test.

Could you guys plz tell me how to export wav to aac/mp4 file? All I get is aac/mov file that's of course is non playable with winamp...

Ok, I found the answer myself (-;
The resulted mp4 plays fine with winamp, and I still can't believe that commercial aac encoder found its way to the world.
Long live Apple!

(That is ISO Compliant, I mean)

In the player, open the wav you're going to encode.

File -> Export

Export box: Movie to MPEG-4 -> click Options

Top box: Select Audio -> choose your bitrate & mono/stereo.

It should now create a .mp4 audio file.

OK, converting files using QT Player is a pain in the ass right now, I wonder what consequences a CD ripper using QT for encoding would cause. I see MP3 like trouble coming from RIAA.

hey !! QT AAC is MAIN profile !!! , not low complexity it's why it is better than LQT

i found that in the "faad winamp plugin property box" of the a qt .aac file

[img]b41ae00ad3[/img]

The QT player indicates that the codec is Low Complexity ???
What's wrong then? Maybe there's something wrong in the MPEG-4
header. Anyways, That's not the source of better sound quality for sure
Main profile of a codec can not reveal much improvement over LC profile.

Edit: After posting my reply, I noticed your attached image. I guess you
have mistaken. Isn't that a screenshot from a Psytell AAC file? Please
check it again I'm using Audio Coding MPEG-4 Audio Plugin for Winamp
and I see LC profile; Even more, the screenshot you've posted has been
taken from a MPEG-2 AAC Plugin; QT doesn't create such a bitstream! ???

Cheers,
Mehrtash

You have some "mistake" here... QT can't compress neither MPEG2 AACs nor AACs in native format (not contained in MP4)... And they ARE low complexity!

you're wrong , it's posssible to extract an AAC file from a MP4 with MP4extract60.exe

you can find it at rarewares

So, there must be a bug in the header of the "extracted" AAC file
The MP4 file that QT creates is LC.

Cheers,
Mehrtash

And they are MPEG4... Something is wrong here, I'll write more after more testing...

Looks like someone has downloaded my 41 tests

These .aac files were generated by:

- stripping RAW (headerless) AAC from .mp4 files generated by quicktime
- making ADTS files out of RAW files with adif2mp4 (ISO/MPEG tool)


I dunno about MAIN profile - anyways, anybody care enough to analyse file called QuickTimeMPEG4Authoring.qtx in windows/system32/quicktime directory, with hex/text viewer, would find out that the debug and error messages are the same as in LiquidAudio 4.x and 5.x files! Only difference is that the files have .c extension and in LiquidAudio the files were .cpp (C++)

One interesting thing, for example:



This is the same message as in LiquidAudio 4.x and 5.x No PNS in this one

One thing that is missing in debug messages is 'wrong predictor on/off initialization' found in liquifier executable - I doubt that prediction is even included in code package in QuickTime.

Regarding Castanets and Fatboy - when I tested QT6 preview , I found weird bugs at 128 Kbps - the codec missed some short blocks, and coded them as long, creating noticeable pre-echo in two or three castanets hits, also fatboy encoding was having same troubles with terrible "chainsaw" effect in one part. Everything else was as same as in LiquidAudio 5.x - except that short block problems.

I suppose the bug has been corrected in the final version - but I havent tested yet.