errors in libfaac

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
14 messages Options
Reply | Threaded
Open this post in threaded view
|

errors in libfaac

vdmsss
Hello.

        I fully expect this is a bug with faac; I am emailing the list to ask  
whether people are aware of it and know any workaround in ffmpeg. I  
converted a few files from .ra to .m4a using simply

        ffmpeg -i test1.ra test1.m4a or ffmpeg -i test1.ra test1.aac

The conversions worked with no warnings whatsoever (full listings  
below), and I am using the most recent versions of ffmpeg, faad and  
faac. However, the output is not complaint. When I try to faad the  
resulting file, or also to ffmpeg-ed it again, I have several errors,  
as eg:

        Gain control not yet implemented
        Channel coupling not yet implemented
        Maximum number of bitstream elements exceeded

For instance:

> faad test1.m4a
>  *********** Ahead Software MPEG-4 AAC Decoder V2.6 ******************
>
>  Build: Jan  4 2008
>  Copyright 2002-2004: Ahead Software AG
>  http://www.audiocoding.com
>  Floating point version
>
>  This program is free software; you can redistribute it and/or modify
>  it under the terms of the GNU General Public License.
>
>  
> **************************************************************************
>
> test1.m4a file info:
>
> LC AAC 899.938 secs, 2 ch, 44100 Hz
>
>   ---------------------
>  | Config:  2 Ch       |
>   ---------------------
>  | Ch |    Position    |
>   ---------------------
>  | 00 | Left front     |
>  | 01 | Right front    |
>   ---------------------
>
> Warning: Gain control not yet implemented
> Warning: Gain control not yet implemented
> Warning: Gain control not yet implemented
> Warning: Gain control not yet implemented
> Warning: Maximum number of bitstream elements exceeded
> Warning: Gain control not yet implemented
> Decoding test1.m4a took:  3.85 sec. 233.59x real-time.
>


Note that test1.m4a plays well in mplayer, QuickTime and iTunes, but  
it crashes my iPod (triggers a reboot), which is how I discovered the  
problem.

Clearly, the same happens if I feed test1.m4a back to ffmpeg: eg

> ffmpeg -i test1.m4a test1.mp3
> FFmpeg version SVN-r11435, Copyright (c) 2000-2008 Fabrice Bellard,  
> et al.
>   configuration: --disable-ffserver --disable-gprof --disable-debug  
> --enable-gpl --enable-libx264 --enable-libxvid --enable-libfaac --
> enable-libfaad --enable-liba52 --enable-libtheora --enable-libamr-wb  
> --enable-libmp3lame --enable-libamr-nb --enable-libgsm --enable-
> pthreads --enable-swscaler --enable-pp --extra-libs=-L/usr/X11R6/lib  
> -lX11 -dylib_file /System/Library/Frameworks/OpenGL.framework/
> Versions/A/Libraries/libGL.dylib:/System/Library/Frameworks/
> OpenGL.framework/Versions/A/Libraries/libGL.dylib --extra-cflags=-O3  
> -mtune=nocona -mmmx -msse -msse2 -msse3 -mfpmath=sse -fomit-frame-
> pointer -pipe -I/sw/include --cpu=nocona
>   libavutil version: 49.6.0
>   libavcodec version: 51.49.0
>   libavformat version: 52.3.0
>   built on Jan  6 2008 18:52:56, gcc: 4.0.1 (Apple Inc. build 5465)
> Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'test1.m4a':
>   Duration: 00:15:00.8, start: 0.000000, bitrate: 66 kb/s
>     Stream #0.0(und): Audio: mpeg4aac, 44100 Hz, stereo
> Output #0, mp3, to 'test1.mp3':
>     Stream #0.0(und): Audio: libmp3lame, 44100 Hz, stereo, 64 kb/s
> Stream mapping:
>   Stream #0.0 -> #0.0
> Press [q] to stop encoding
> [mpeg4aac @ 0x528930]faac: frame decoding failed: Gain control not  
> yet implemented
> Error while decoding stream #0.0
> [mpeg4aac @ 0x528930]faac: frame decoding failed: Gain control not  
> yet implemented
> Error while decoding stream #0.0
> [mpeg4aac @ 0x528930]faac: frame decoding failed: Gain control not  
> yet implemented
> Error while decoding stream #0.0
> [mpeg4aac @ 0x528930]faac: frame decoding failed: Gain control not  
> yet implemented
> Error while decoding stream #0.0
> [mpeg4aac @ 0x528930]faac: frame decoding failed: Maximum number of  
> bitstream elements exceeded
> Error while decoding stream #0.0
> [mpeg4aac @ 0x528930]faac: frame decoding failed: Gain control not  
> yet implemented
> Error while decoding stream #0.0
> size=    7037kB time=900.7 bitrate=  64.0kbits/s
> video:0kB audio:7037kB global headers:0kB muxing overhead 0.000430%
>


Very similar problems with the aac encoding. Interesting fact is that  
if I first encode with libmp3lame and then with libfaac (full logs  
below),

         ffmpeg -i test1.ra test1.mp3 ; ffmpeg -i test1.mp3 test2.mp4

then the resulting files are OK, that is faad (and ffmpeg) decodes  
without any warnings.

So is the lesson here simply: never to go straight from cook-decoding  
to faac-encoding?

regards

=====
LOGS:

> ffmpeg -i test1.ra test1.m4a
> FFmpeg version SVN-r11435, Copyright (c) 2000-2008 Fabrice Bellard,  
> et al.
>   configuration: --disable-ffserver --disable-gprof --disable-debug  
> --enable-gpl --enable-libx264 --enable-libxvid --enable-libfaac --
> enable-libfaad --enable-liba52 --enable-libtheora --enable-libamr-wb  
> --enable-libmp3lame --enable-libamr-nb --enable-libgsm --enable-
> pthreads --enable-swscaler --enable-pp --extra-libs=-L/usr/X11R6/lib  
> -lX11 -dylib_file /System/Library/Frameworks/OpenGL.framework/
> Versions/A/Libraries/libGL.dylib:/System/Library/Frameworks/
> OpenGL.framework/Versions/A/Libraries/libGL.dylib --extra-cflags=-O3  
> -mtune=nocona -mmmx -msse -msse2 -msse3 -mfpmath=sse -fomit-frame-
> pointer -pipe -I/sw/include --cpu=nocona
>   libavutil version: 49.6.0
>   libavcodec version: 51.49.0
>   libavformat version: 52.3.0
>   built on Jan  6 2008 18:52:56, gcc: 4.0.1 (Apple Inc. build 5465)
> Input #0, rm, from 'test1.ra':
>   Duration: 00:15:01.1, start: 0.000000, bitrate: 44 kb/s
>     Stream #0.0: Audio: cook, 44100 Hz, stereo, 44 kb/s
> Output #0, mp4, to 'test1.m4a':
>     Stream #0.0: Audio: libfaac, 44100 Hz, stereo, 64 kb/s
> Stream mapping:
>   Stream #0.0 -> #0.0
> Press [q] to stop encoding
> size=    7337kB time=900.8 bitrate=  66.7kbits/s
> video:0kB audio:7033kB global headers:0kB muxing overhead 4.317598%

=========


> ffmpeg -i test1.ra test1.mp3; ffmpeg -i test1.mp3 test2.m4a
> FFmpeg version SVN-r11435, Copyright (c) 2000-2008 Fabrice Bellard,  
> et al.
>   configuration: --disable-ffserver --disable-gprof --disable-debug  
> --enable-gpl --enable-libx264 --enable-libxvid --enable-libfaac --
> enable-libfaad --enable-liba52 --enable-libtheora --enable-libamr-wb  
> --enable-libmp3lame --enable-libamr-nb --enable-libgsm --enable-
> pthreads --enable-swscaler --enable-pp --extra-libs=-L/usr/X11R6/lib  
> -lX11 -dylib_file /System/Library/Frameworks/OpenGL.framework/
> Versions/A/Libraries/libGL.dylib:/System/Library/Frameworks/
> OpenGL.framework/Versions/A/Libraries/libGL.dylib --extra-cflags=-O3  
> -mtune=nocona -mmmx -msse -msse2 -msse3 -mfpmath=sse -fomit-frame-
> pointer -pipe -I/sw/include --cpu=nocona
>   libavutil version: 49.6.0
>   libavcodec version: 51.49.0
>   libavformat version: 52.3.0
>   built on Jan  6 2008 18:52:56, gcc: 4.0.1 (Apple Inc. build 5465)
> Input #0, rm, from 'test1.ra':
>   Duration: 00:15:01.1, start: 0.000000, bitrate: 44 kb/s
>     Stream #0.0: Audio: cook, 44100 Hz, stereo, 44 kb/s
> Output #0, mp3, to 'test1.mp3':
>     Stream #0.0: Audio: libmp3lame, 44100 Hz, stereo, 64 kb/s
> Stream mapping:
>   Stream #0.0 -> #0.0
> Press [q] to stop encoding
> size=    7039kB time=900.9 bitrate=  64.0kbits/s
> video:0kB audio:7039kB global headers:0kB muxing overhead 0.000430%
> FFmpeg version SVN-r11435, Copyright (c) 2000-2008 Fabrice Bellard,  
> et al.
>   configuration: --disable-ffserver --disable-gprof --disable-debug  
> --enable-gpl --enable-libx264 --enable-libxvid --enable-libfaac --
> enable-libfaad --enable-liba52 --enable-libtheora --enable-libamr-wb  
> --enable-libmp3lame --enable-libamr-nb --enable-libgsm --enable-
> pthreads --enable-swscaler --enable-pp --extra-libs=-L/usr/X11R6/lib  
> -lX11 -dylib_file /System/Library/Frameworks/OpenGL.framework/
> Versions/A/Libraries/libGL.dylib:/System/Library/Frameworks/
> OpenGL.framework/Versions/A/Libraries/libGL.dylib --extra-cflags=-O3  
> -mtune=nocona -mmmx -msse -msse2 -msse3 -mfpmath=sse -fomit-frame-
> pointer -pipe -I/sw/include --cpu=nocona
>   libavutil version: 49.6.0
>   libavcodec version: 51.49.0
>   libavformat version: 52.3.0
>   built on Jan  6 2008 18:52:56, gcc: 4.0.1 (Apple Inc. build 5465)
> Input #0, mp3, from 'test1.mp3':
>   Duration: 00:15:00.9, start: 0.000000, bitrate: 64 kb/s
>     Stream #0.0: Audio: mp3, 44100 Hz, stereo, 64 kb/s
> Output #0, mp4, to 'test2.m4a':
>     Stream #0.0: Audio: libfaac, 44100 Hz, stereo, 64 kb/s
> Stream mapping:
>   Stream #0.0 -> #0.0
> Press [q] to stop encoding
> size=    7339kB time=900.9 bitrate=  66.7kbits/s
> video:0kB audio:7036kB global headers:0kB muxing overhead 4.316233%
>


=========

> faad test2.m4a
>  *********** Ahead Software MPEG-4 AAC Decoder V2.6 ******************
>
>  Build: Jan  4 2008
>  Copyright 2002-2004: Ahead Software AG
>  http://www.audiocoding.com
>  Floating point version
>
>  This program is free software; you can redistribute it and/or modify
>  it under the terms of the GNU General Public License.
>
>  
> **************************************************************************
>
> test2.m4a file info:
>
> LC AAC 899.985 secs, 2 ch, 44100 Hz
>
>   ---------------------
>  | Config:  2 Ch       |
>   ---------------------
>  | Ch |    Position    |
>   ---------------------
>  | 00 | Left front     |
>  | 01 | Right front    |
>   ---------------------
>
> Decoding test1.m4a took:  3.87 sec. 232.34x real-time.
>
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Benjamin Larsson-2
vdmsss wrote:

> Hello.
>
> I fully expect this is a bug with faac; I am emailing the list to ask  
> whether people are aware of it and know any workaround in ffmpeg. I  
> converted a few files from .ra to .m4a using simply
>
> ffmpeg -i test1.ra test1.m4a or ffmpeg -i test1.ra test1.aac
>
> The conversions worked with no warnings whatsoever (full listings  
> below), and I am using the most recent versions of ffmpeg, faad and  
> faac. However, the output is not complaint. When I try to faad the  
> resulting file, or also to ffmpeg-ed it again, I have several errors,  
> as eg:
>
> Gain control not yet implemented
> Channel coupling not yet implemented
> Maximum number of bitstream elements exceeded
>
>  

Someone mentioned that faac sometimes produces corrupt encodings. And as
such it can create bad streams randomly. And I have no better idea then
using the latest release of faac.

MvH
Benjamin Larsson


_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Baptiste Coudurier
Hi,

Benjamin Larsson wrote:

> vdmsss wrote:
>> Hello.
>>
>> I fully expect this is a bug with faac; I am emailing the list to ask  
>> whether people are aware of it and know any workaround in ffmpeg. I  
>> converted a few files from .ra to .m4a using simply
>>
>> ffmpeg -i test1.ra test1.m4a or ffmpeg -i test1.ra test1.aac
>>
>> The conversions worked with no warnings whatsoever (full listings  
>> below), and I am using the most recent versions of ffmpeg, faad and  
>> faac. However, the output is not complaint. When I try to faad the  
>> resulting file, or also to ffmpeg-ed it again, I have several errors,  
>> as eg:
>>
>> Gain control not yet implemented
>> Channel coupling not yet implemented
>> Maximum number of bitstream elements exceeded
>>
>>  
>
> Someone mentioned that faac sometimes produces corrupt encodings. And as
> such it can create bad streams randomly. And I have no better idea then
> using the latest release of faac.
>

Well IMHO this is a serious issue, and maybe we could address it,
faad/faac becomes more and more unreliable it seems. We should have soon
a replacement for faad but faac should be LGPL and replacement is
scheduled...

Let's resurect the thread about glueing code ?

--
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

vdmsss
On 7 Jan 2008, at 14:29, Baptiste Coudurier wrote:
>> Someone mentioned that faac sometimes produces corrupt encodings.  
>> And as
>> such it can create bad streams randomly. And I have no better idea  
>> then
>> using the latest release of faac.
>>

Please notice that I am using CVS versions of faad and faac compiled  
only a few days ago.

> Well IMHO this is a serious issue, and maybe we could address it,
> faad/faac becomes more and more unreliable it seems. We should have  
> soon
> a replacement for faad but faac should be LGPL and replacement is
> scheduled...

It does seem serious. AAC is an important codec, and libfaac is the  
reference encoder, and the only one available in ffmpeg. I am  
surprised that not only it appears to generate code that faad can't  
decompress ("not yet implemented" features), it also seems to breach  
the container/profile rules ("maximum number of bitstream elements  
exceeded"). And it gives no warning whatsoever; on this occasion my  
test file was a particularly simple 64kbit speech-only radio  
programme: I am now wondering how many of my m4a are indeed corrupt.


--




_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Rich Felker
On Mon, Jan 07, 2008 at 08:49:23PM +0000, vdmsss wrote:
> It does seem serious. AAC is an important codec, and libfaac is the  

AAC is an _extremely_ bad codec that should only be used in situations
where it's mandated by the need for compatibility with a particular
device. Vorbis has better quality with roughly 1/10 of the complexity,
as do plenty of other codecs.

Rich
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Christian Ebert
* Rich Felker on Tuesday, January 08, 2008 at 00:29:39 -0500
> On Mon, Jan 07, 2008 at 08:49:23PM +0000, vdmsss wrote:
>> It does seem serious. AAC is an important codec, and libfaac is the  
>
> AAC is an _extremely_ bad codec that should only be used in situations
> where it's mandated by the need for compatibility with a particular
> device.

As in for DVDs played by hardware players?

> Vorbis has better quality with roughly 1/10 of the complexity,
> as do plenty of other codecs.

AFAIK, for DVDs mp2 would be the only alternative?

c
--
_B A U S T E L L E N_ lesen! --->> <http://www.blacktrash.org/baustellen.html>
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Nico Sabbi
Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
> As in for DVDs played by hardware players?
>
> > Vorbis has better quality with roughly 1/10 of the complexity,
> > as do plenty of other codecs.
>
> AFAIK, for DVDs mp2 would be the only alternative?
>
> c

mp2, ac3, lpcm and dts are the only formats allowed in dvd;
aac was (fortunately) never included
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Rich Felker
In reply to this post by Christian Ebert
On Tue, Jan 08, 2008 at 12:15:12PM +0100, Christian Ebert wrote:
> * Rich Felker on Tuesday, January 08, 2008 at 00:29:39 -0500
> > On Mon, Jan 07, 2008 at 08:49:23PM +0000, vdmsss wrote:
> >> It does seem serious. AAC is an important codec, and libfaac is the  
> >
> > AAC is an _extremely_ bad codec that should only be used in situations
> > where it's mandated by the need for compatibility with a particular
> > device.
>
> As in for DVDs played by hardware players?

No, only iPods or Sony PSP or something like that; I forget which ones
exactly are the ones that require AAC...

Rich
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Christian Ebert
In reply to this post by Nico Sabbi
* Nico Sabbi on Tuesday, January 08, 2008 at 12:19:31 +0100
> Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
>> As in for DVDs played by hardware players?
>>
>>> Vorbis has better quality with roughly 1/10 of the complexity,
>>> as do plenty of other codecs.
>>
>> AFAIK, for DVDs mp2 would be the only alternative?
>
> mp2, ac3,

arrgh, I always get ac3 and aac muddled up, thanks for
correcting.

> lpcm and dts are the only formats allowed in dvd;

and which one would you say is best of those four?

> aac was (fortunately) never included

right.

c
--
_B A U S T E L L E N_ lesen! --->> <http://www.blacktrash.org/baustellen.html>
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Nico Sabbi
Il Tuesday 08 January 2008 12:59:06 Christian Ebert ha scritto:

> * Nico Sabbi on Tuesday, January 08, 2008 at 12:19:31 +0100
>
> > Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
> >> As in for DVDs played by hardware players?
> >>
> >>> Vorbis has better quality with roughly 1/10 of the complexity,
> >>> as do plenty of other codecs.
> >>
> >> AFAIK, for DVDs mp2 would be the only alternative?
> >
> > mp2, ac3,
>
> arrgh, I always get ac3 and aac muddled up, thanks for
> correcting.
>
> > lpcm and dts are the only formats allowed in dvd;
>
> and which one would you say is best of those four?
>

except lpcm (uncompressed) the ac3 encoder should sound better
than the mp2 counterpart (that sounds really bad)
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Baptiste Coudurier
Hi

Nico Sabbi wrote:

> Il Tuesday 08 January 2008 12:59:06 Christian Ebert ha scritto:
>> * Nico Sabbi on Tuesday, January 08, 2008 at 12:19:31 +0100
>>
>>> Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
>>>> As in for DVDs played by hardware players?
>>>>
>>>>> Vorbis has better quality with roughly 1/10 of the complexity,
>>>>> as do plenty of other codecs.
>>>> AFAIK, for DVDs mp2 would be the only alternative?
>>> mp2, ac3,
>> arrgh, I always get ac3 and aac muddled up, thanks for
>> correcting.
>>
>>> lpcm and dts are the only formats allowed in dvd;
>> and which one would you say is best of those four?
>>
>
> except lpcm (uncompressed) the ac3 encoder should sound better
> than the mp2 counterpart (that sounds really bad)

Maybe at 64k twolame/toolame using psychoacoustic model is better and
nothing really great either, if you disable psychoacoustic (-P -1) then
sound quality is the same.

But 128k and higher, quality is likely equivalent. Therefore for a DVD,
ff mp2 encoder is good.

--
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

vdmsss
In reply to this post by Nico Sabbi
On 8 Jan 2008, at 11:19, Nico Sabbi wrote:

> Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
>> As in for DVDs played by hardware players?
>>
>>> Vorbis has better quality with roughly 1/10 of the complexity,
>>> as do plenty of other codecs.
>>
>> AFAIK, for DVDs mp2 would be the only alternative?
>>
>> c
>
> mp2, ac3, lpcm and dts are the only formats allowed in dvd;
> aac was (fortunately) never included

regardless of the (de)merits of aac (in truth I haven't heard many bad  
things, but I'm happy to believe you), it doesn't look like vorbis is  
supported in mp4, and people will want ipod-compatible codecs and  
containers. So, the viable alternatives to guarantee high  
compatibility remain mp2/ac3 for DVDs and mp3 for the rest. Yet, I  
feel it's a shame if ffmpeg/mplayer were not to have a high quality  
aac option. I thought it was a part of SOC to develop an aac decoder  
for ffmpeg, don't know about encoders.

--
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Rich Felker
In reply to this post by Baptiste Coudurier
On Tue, Jan 08, 2008 at 01:37:37PM +0100, Baptiste Coudurier wrote:

> Hi
>
> Nico Sabbi wrote:
> > Il Tuesday 08 January 2008 12:59:06 Christian Ebert ha scritto:
> >> * Nico Sabbi on Tuesday, January 08, 2008 at 12:19:31 +0100
> >>
> >>> Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
> >>>> As in for DVDs played by hardware players?
> >>>>
> >>>>> Vorbis has better quality with roughly 1/10 of the complexity,
> >>>>> as do plenty of other codecs.
> >>>> AFAIK, for DVDs mp2 would be the only alternative?
> >>> mp2, ac3,
> >> arrgh, I always get ac3 and aac muddled up, thanks for
> >> correcting.
> >>
> >>> lpcm and dts are the only formats allowed in dvd;
> >> and which one would you say is best of those four?
> >>
> >
> > except lpcm (uncompressed) the ac3 encoder should sound better
> > than the mp2 counterpart (that sounds really bad)
>
> Maybe at 64k twolame/toolame using psychoacoustic model is better and
> nothing really great either, if you disable psychoacoustic (-P -1) then
> sound quality is the same.
>
> But 128k and higher, quality is likely equivalent. Therefore for a DVD,
> ff mp2 encoder is good.

IIRC the only formats _required_ to be supported for NTSC DVD are lpcm
and ac3. If you use mp2 or dts your disc might not be playable on all
players. So ac3 is the best to use.

If I'm not mistaken, PAL DVD requires mp2 to be supported as well, so
you have more choices.

Rich
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user
Reply | Threaded
Open this post in threaded view
|

Re: errors in libfaac

Rich Felker
In reply to this post by vdmsss
On Tue, Jan 08, 2008 at 02:16:16PM +0000, vdmsss wrote:

> On 8 Jan 2008, at 11:19, Nico Sabbi wrote:
> > Il Tuesday 08 January 2008 12:15:12 Christian Ebert ha scritto:
> >> As in for DVDs played by hardware players?
> >>
> >>> Vorbis has better quality with roughly 1/10 of the complexity,
> >>> as do plenty of other codecs.
> >>
> >> AFAIK, for DVDs mp2 would be the only alternative?
> >>
> >> c
> >
> > mp2, ac3, lpcm and dts are the only formats allowed in dvd;
> > aac was (fortunately) never included
>
> regardless of the (de)merits of aac (in truth I haven't heard many bad  
> things, but I'm happy to believe you), it doesn't look like vorbis is  
> supported in mp4, and people will want ipod-compatible codecs and  
> containers. So, the viable alternatives to guarantee high  
> compatibility remain mp2/ac3 for DVDs and mp3 for the rest. Yet, I  
> feel it's a shame if ffmpeg/mplayer were not to have a high quality  
> aac option. I thought it was a part of SOC to develop an aac decoder  
> for ffmpeg, don't know about encoders.

Decoder is highly desirable for any codec that's in widespread use.
Encoder is only desirable if the codec is actually worth using,
either for quality or for compatibility.

Rich
_______________________________________________
ffmpeg-user mailing list
[hidden email]
http://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-user