qtaacenc: a command-line QuickTime AAC encoder for Windows Options

[wkmax]

Hello.

I have one question about VBR encoding in QuickTime. I was searching with Google but have not found detailed explanation of Constrained VBR mode.

Is it better or worse than True VBR? . Maybe it gives higher quality than tvbr only at maximum quality settings (--cvbr 320 VS --tvbr 127) but at lower bitrates I can get better quality with true vbr?

This link might be useful for you:

developer.apple.com/mac/library/technotes/tn2008/tn2237.html
(particularly "encoding properties in detail" section)

Regards

This post has been edited by wkmax: Feb 15 2010, 19:34

[Steve Forte Rio]

rpp3po, wkmax, thank you!

Another question: why QT AAC Encoder does not allow to encode at high bitrates (>320 kbps) in CBR mode while in VBR mode it supports frames up to 500 kbps o.0

This post has been edited by Steve Forte Rio: Feb 16 2010, 17:41

[rpp3po]

It's impolite to answer with a counter question, but why should it? It makes sense for VBR in case it encounters super complex passages, but why would one want to force an encoder to such high rates? I even question the sanity of forcing a modern implementation as QT or Nero to not drop below 320 kbit/s.

[Steve Forte Rio]

I understand what are you talking about, but I know some people who do not trust the algorithms of choosing bitrates for each frame (complexity detecting algorithms maybe), especially in True VBR mode... And they are not sure of VBR mode safety. And I think that CBR mode with high bitrates will be the safest for the quality despite the fact that it is waste of file size

(sorry for my very bad English)

[rpp3po]

IMHO it is a developer's responsibility to expose reasonable settings for his product, and that does not extend to satisfying feelings as unfounded paranoia. Exposing insane bitrate settings may raise the impression that there might actually be a need for them. If you can present a sample, that would prove otherwise, feel free to share it. People who cannot get over their distrust against lossy compression should go for lossless, anyway, because even extreme bitrates cannot guarantee transparency for 100% of music.

PS BTW, the complexity of the TVBR encoder is actually lower than that of CVBR, ABR, and CBR. What you call "complexity detection" is required for all stages, but the latter must calculate a trade-off against additional constraints.

This post has been edited by rpp3po: Feb 16 2010, 19:11

[Steve Forte Rio]

Thanks for your answers, it's all as I expected

best regards

[nao]

Important update

Though I said HE-AAC encoder was not available in the Windows version of QT, I've found a way to access the HE-AAC encoder. So, now the latest version has an ability to create HE-AAC files.

--he option enables the HE-AAC encoder. Note that the true vbr mode is not available in this encoder.

Enjoy!

[Steve Forte Rio]

Cool! Thank you.

And does QuickTime have Parametric Stereo technology support ?

[nao]

No. HE-AAC v1 only.

[tedgo]

Thank you very much .
Is there any "hardcoded" bitrate limit the HE-profile can be used with?

[nao]

The limit cahnges according to the samplerate and the number of channel. For example, 80kbps for 44.1kHz, stereo input.

[tedgo]

Thanks, i tried it now.
Setting the bitrate higher than 80kbps encodes at 80kbps.

This post has been edited by tedgo: Feb 17 2010, 18:03

[wkmax]

Important update

Though I said HE-AAC encoder was not available in the Windows version of QT, I've found a way to access the HE-AAC encoder. So, now the latest version has an ability to create HE-AAC files.

--he option enables the HE-AAC encoder. Note that the true vbr mode is not available in this encoder.

Enjoy!

Excellent!!!, works fine.
Thanks again nao

This post has been edited by wkmax: Feb 18 2010, 09:46

[Steve Forte Rio]

Another questions:

What about 32-bit floating-point input support?

And is it possible to realize two-pass encoding in QuickTime?

[nao]

32-bit fp input is already supported. 2-pass encoding is not possible for AAC.

[rpp3po]

It is, Nero supports it. But its kind of senseless. During the good old divx times it made sense to use 2-pass to exactly squeeze a video onto 2 x 700 MB CDRs. 2-pass was the perfect tool, to use up all available storage space up to the last MB. But for AAC it's basically completely senseless.

This post has been edited by rpp3po: Feb 18 2010, 15:42

[nao]

It is, Nero supports it.

I should have said "not possible in QuickTime" to be correct.

[Steve Forte Rio]

32-bit fp input is already supported.

This is great! Big thank you!

[b66pak]

@nao how do you explain this?


using qtaacenc-20100212 on a 121 minutes 6ch wav i get 538,097,995 bytes and using qtaacenc-20100218 with the same settings a get 558,822,574 bytes!


encoding line:

CODE

bepipe --script "import(^rawav.avs^)" | qtaacenc-20100212 --ignorelength --samplerate keep --highest --tvbr 127 - audio.6ch.q127.qtaacenc-20100212.m4a

result:

CODE

538,097,995 bytes



General
Complete name                    : audio.6ch.q127.qtaacenc-20100212.m4a
Format                           : MPEG-4
Format profile                   : Base Media / Version 2
Codec ID                         : mp42
File size                        : 513 MiB
Duration                         : 2h 1mn
Overall bit rate                 : 589 Kbps
Encoded date                     : UTC 2010-02-18 19:10:27
Tagged date                      : UTC 2010-02-18 19:11:02
Writing application              : qtaacenc 20100212, QuickTime 7.6.5
Encoding Params                  : (Binary)

Audio
ID                               : 1
Format                           : AAC
Format/Info                      : Advanced Audio Codec
Format version                   : Version 4
Format profile                   : LC
Format settings, SBR             : No
Codec ID                         : 40
Duration                         : 2h 1mn
Bit rate mode                    : Variable
Bit rate                         : 576 Kbps
Maximum bit rate                 : 847 Kbps
Channel(s)                       : 6 channels
Channel positions                : Front: L C R, Rear: L R, LFE
Sampling rate                    : 48.0 KHz
Stream size                      : 511 MiB (100%)
Language                         : English
Encoded date                     : UTC 2010-02-18 19:10:19
Tagged date                      : UTC 2010-02-18 19:11:02

encoding line:

CODE

bepipe --script "import(^rawav.avs^)" | qtaacenc-20100218 --ignorelength --samplerate keep --highest --tvbr 127 - audio.6ch.q127.qtaacenc-20100218.m4a

result:

CODE

558,822,574 bytes



General
Complete name                    : audio.6ch.q127.qtaacenc-20100218.m4a
Format                           : MPEG-4
Format profile                   : Base Media / Version 2
Codec ID                         : mp42
File size                        : 533 MiB
Duration                         : 2h 1mn
Overall bit rate                 : 612 Kbps
Encoded date                     : UTC 2010-02-18 18:35:26
Tagged date                      : UTC 2010-02-18 18:36:02
Writing application              : qtaacenc 20100218, QuickTime 7.6.5
Encoding Params                  : (Binary)

Audio
ID                               : 1
Format                           : AAC
Format/Info                      : Advanced Audio Codec
Format version                   : Version 4
Format profile                   : LC
Format settings, SBR             : No
Codec ID                         : 40
Duration                         : 2h 1mn
Bit rate mode                    : Variable
Bit rate                         : 609 Kbps
Maximum bit rate                 : 962 Kbps
Channel(s)                       : 6 channels
Channel positions                : Front: L C R, Rear: L R, LFE
Sampling rate                    : 48.0 KHz
Stream size                      : 531 MiB (100%)
Language                         : English
Encoded date                     : UTC 2010-02-18 18:35:18
Tagged date                      : UTC 2010-02-18 18:36:02

best regards
_

[nao]

@nao how do you explain this?

Probably the input file has no channel order info (dwChannelMask in WAVEFORMATEXTENSIBLE). The default order was C L R LFE Ls Rs in the version 20100212, but I changed it to L R C LFE Ls Rs in the version 20100217.

[b66pak]

wrong channel order...you should add to the log file that you fix this!...what is the max rate for multichannel audio?...it has real mutichannel support (2.1, 4.0 etc) or 5.1 ONLY?
_

[rpp3po]

If your source file doesn't include any channel mapping information, there is no such thing as a wrong channel order after conversion. For cases where this information is missing, there is a default ordering proposed by Microsoft, 20100217 uses exactly that.

This post has been edited by rpp3po: Feb 18 2010, 22:11

[nao]

what is the max rate for multichannel audio?...it has real mutichannel support (2.1, 4.0 etc) or 5.1 ONLY?

Maximum supported samplerate is 48kHz. Supported multichannel formats are:

• LC-AAC
  
  • Quadraphonic (L R Ls Rs)
  • 4.0ch (L R C Cs)
  • 5.0ch (L R C Ls Rs)
  • 5.1ch (L R C LFE Ls Rs)
  • 6.0ch (L R C Ls Rs Cs)
  • 6.1ch (L R C LFE Ls Rs Cs)
  • 7.0ch (L R C Ls Rs Rls Rrs)
  • 7.1ch (L R C LFE Ls Rs Lc Rc)
• HE-AAC
  
  • Quadraphonic (L R Ls Rs)
  • 5.1ch (L R C LFE Ls Rs)
  • 7.1ch (L R C LFE Ls Rs Lc Rc)

The result in case the channel order info is missing may be inconsistent.

[b66pak]

what about max bitrate for multichannel audio?
_

[nao]

I forgot to mention, but 8.0ch is also supported. The bitrate limits (@48kHz) are:

• LC-AAC
  
  • Quadraphonic : 640 kbps
  • 4.0ch : 640 kbps
  • 5.0ch : 768 kbps
  • 5.1ch : 768 kbps
  • 6.0ch : 960 kbps
  • 6.1ch : 960 kbps
  • 7.0ch : 960 kbps
  • 7.1ch : 960 kbps
  • 8.0ch (L R C Ls Rs Cs Rls Rrs) : 1280 kbps
• HE-AAC
  
  • Quadraphonic : 160 kbps
  • 5.1ch : 192 kbps
  • 7.1ch : 256 kbps

The actual limit changes according to the input samplerate.