Aften 0.0.8 is out

[DarkAvenger]

btw, what are the pros and the cons about cmake vs make ?


Wrong question. ;) CMake uses make depending on your platform. CMake is a substitution for configure. Its use is portability. It has support to build Makefiles/project files for various Compilers/IDEs.

So as dev, we just have to maintain CMake - nothing else and it is a lot easier to use than eg autotools or a manual script as Justin used. And as user/distributor you can generate a native project file for his needs (as long as cmake supports it).


A small how-to could be great too coz I've several compilers (gcc within MinGW, MSVC6, MSVC8).

For windows you need cureently CMake built from cvs, as the stable version has a bug with custom languages support.
For that you need (I used mingw to compile cmake):

- mingw (including c++ compiler) ;)
- cvs (best added to path)
- cmake from cvs (see cmake.org download page)

Go into Cmake's root dir type ./bootstrap (or alike) then make and wait. make install will put it into program files and should also add it to path.

Now for aften. I think aou already have the svn version.

- Create a dir (eg def), change into def.

Depending on the compiler you want to use, you need to specify the generator.

So for MSYS/MinGW (in dir def):

- cmake -G "MSYS Makefiles" ..
- make

(You need to manually strip the binaries.)

For MSVC you could try out the various VC generator. I used the Visual C++ 2005 Express edition w/o IDE. for that you need the Platform SDK as well. To get it compiled, start the correct platform sdk env then in the shell start the vc env setting bat

Now to compile aften (using Nmake):

- create a dir (eg vc) and change into vc
- cmake -G "NMake Makefiles" ..
- nmake

That should do it. (Just to note: I don't know if the generator names above are correct. Just type cmake and it wil list available generators. There is also an IDE version of cmake on every platform, eg ccmake for Linux. But I usually prefer command line.)

[Kurtnoise]

Thanks but...


cmake -G "MSYS Makefiles" ..
-- Check for working C compiler: c:/MinGW/bin/gcc.exe
-- Check for working C compiler: c:/MinGW/bin/gcc.exe -- works
-- Check size of void*
-- Check size of void* - done
Could not detect machine type
-- Assuming i386 machine
-- Performing Test HAVE_64BITS
-- Performing Test HAVE_64BITS - Failed
-- Performing Test HAVE_FLAG_STD=GNU99
-- Performing Test HAVE_FLAG_STD=GNU99 - Success
-- Performing Test HAVE_FLAG_WDISABLED_OPTIMIZATION
-- Performing Test HAVE_FLAG_WDISABLED_OPTIMIZATION - Success
-- Performing Test HAVE_FLAG_WFLOAT_EQUAL
-- Performing Test HAVE_FLAG_WFLOAT_EQUAL - Success
-- Performing Test HAVE_FLAG_WBAD_FUNCTION_CAST
-- Performing Test HAVE_FLAG_WBAD_FUNCTION_CAST - Success
-- Performing Test HAVE_FLAG_WDECLARATION_AFTER_STATEMENT
-- Performing Test HAVE_FLAG_WDECLARATION_AFTER_STATEMENT - Success
-- Performing Test HAVE_FLAG_WEXTRA
-- Performing Test HAVE_FLAG_WEXTRA - Success
-- Performing Test HAVE_FLAG_WNO_SWITCH
-- Performing Test HAVE_FLAG_WNO_SWITCH - Success
-- Check if the system is big endian
-- Check if the system is big endian - little endian
-- Looking for inttypes.h
-- Looking for inttypes.h - found
-- Looking for byteswap.h
-- Looking for byteswap.h - not found
-- Performing Test HAVE_SSE
-- Performing Test HAVE_SSE - Success
-- Performing Test HAVE_SSE3
-- Performing Test HAVE_SSE3 - Success
-- Performing Test HAVE_MM_MALLOC
-- Performing Test HAVE_MM_MALLOC - Success
-- Performing Test HAVE_CASTSI128
-- Performing Test HAVE_CASTSI128 - Failed
-- Using YASM/NASM
-- Performing Test HAVE_NASM_VISIBILITY
-- Performing Test HAVE_NASM_VISIBILITY - Failure
-- Writing config.h
-- Configuring done
-- Generating done
-- Build files have been written to: c:/temp/aften/default

HP_2@LIONEL /c/temp/aften/default
$ make
Scanning dependencies of target aften_static
[ 4%] Building C object CMakeFiles/aften_static.dir/libaften/a52enc.obj
[ 9%] Building C object CMakeFiles/aften_static.dir/libaften/bitalloc.obj
[ 14%] Building C object CMakeFiles/aften_static.dir/libaften/bitio.obj
[ 19%] Building C object CMakeFiles/aften_static.dir/libaften/crc.obj
[ 23%] Building C object CMakeFiles/aften_static.dir/libaften/dynrng.obj
[ 28%] Building C object CMakeFiles/aften_static.dir/libaften/window.obj
[ 33%] Building C object CMakeFiles/aften_static.dir/libaften/mdct.obj
[ 38%] Building C object CMakeFiles/aften_static.dir/libaften/exponent.obj
[ 42%] Building C object CMakeFiles/aften_static.dir/libaften/filter.obj
[ 47%] Building C object CMakeFiles/aften_static.dir/libaften/util.obj
[ 52%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_cpu_caps.obj
[ 57%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_sse_mdct_dummy.obj
[ 61%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_sse_mdct_common_init.obj
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c: In function `sse_mdct_ctx_init':
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:143: warning: 'XMM0' might be used uninitialized in this function
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:143: warning: 'XMM2' might be used uninitialized in this function
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:169: warning: 'XMM0' might be used uninitialized in this function
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:169: warning: 'XMM2' might be used uninitialized in this function
[ 66%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_sse3_mdct_dummy.obj
In file included from c:/temp/aften/libaften/x86/x86_sse3_mdct_dummy.c:2:
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c: In function `mdct_bitreverse':
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:430: warning: implicit declaration of function `_mm_castsi128_ps'
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:430: error: incompatible types in assignment
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:431: error: incompatible types in assignment
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:432: error: incompatible types in assignment
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:433: error: incompatible types in assignment
make[2]: *** [CMakeFiles/aften_static.dir/libaften/x86/x86_sse3_mdct_dummy.obj] Error 1
make[1]: *** [CMakeFiles/aften_static.dir/all] Error 2
make: *** [all] Error 2

compilation failed with minGW. :( Any idea ?

[DarkAvenger]

Thanks for the report. It must be some bug in the code (probably a typo), I'll check, as cmake detects that your compiler doesn't have the mentioned cast function, but tries to use it anyway. Interesting, that it works with my mingw version. :-/


Uhm, could you please post the contents of config.h?

Forget it, I found the bug. When I refactored the code I forgot to implicitly include config.h in the simd support routine, thus no fallback got ever activated... So please svn up for rev 213 and try again.

[Kurtnoise]

Sounds better but not perfect...:)

[ 61%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_sse_mdct_common_init.obj
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c: In function `sse_mdct_ctx_init':
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:143: warning: 'XMM0' might be used uninitialized in this function
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:143: warning: 'XMM2' might be used uninitialized in this function
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:169: warning: 'XMM0' might be used uninitialized in this function
c:/temp/aften/libaften/x86/x86_sse_mdct_common_init.c:169: warning: 'XMM2' might be used uninitialized in this function
[ 66%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_sse3_mdct_dummy.obj
[ 71%] Building ASM object CMakeFiles/aften_static.dir/libaften/x86/x86_cpu_caps_detect.obj
Linking C static library libaften_static.a
[ 71%] Built target aften_static
Scanning dependencies of target aften_wav
[ 76%] Building C object CMakeFiles/aften_wav.dir/aften/wav.obj
Linking C static library libaften_wav.a
[ 76%] Built target aften_wav
Scanning dependencies of target aften_exe
[ 80%] Building C object CMakeFiles/aften_exe.dir/aften/aften.obj
c:/temp/aften/aften/aften.c: In function `print_help':
c:/temp/aften/aften/aften.c:151: warning: string length `4456' is greater than the length `4095' ISO C99 compilers are required to support
Linking C executable aften.exe
c:/temp/aften/default/libaften_static.a(x86_cpu_caps.obj):x86_cpu_caps.c:(.text+0x41): undefined reference to `_alDetectx86CPUCaps'
collect2: ld returned 1 exit status
make[2]: *** [aften.exe] Error 1
make[1]: *** [CMakeFiles/aften_exe.dir/all] Error 2
make: *** [all] Error 2

[DarkAvenger]

I think you used nasm as assembler, right? I think I used yasm. So I'll check outwhat goes wrong with nasm...

OK, should be fixed now.

[Kurtnoise]

Indeed...it works fine now. Thank you very much.

http://kurtnoise.free.fr/index.php?dir=Aften/&file=aften-0.05_rev214.zip

[Kurtnoise]

Hi,

I've made a benchmark regarding encoding speed with a wav sample (44100 Hz - 2Ch - 0:48:26.450 - 488.97 MB) encoded @448kbps :

http://img241.imageshack.us/img241/8757/aftenbenchbi3.png

Note:
OS: WinXP SP2 AMD Athlon 64 3200+ / 2GO Ram
SoftEncode: version 1.0 build 19.
Aften_SSE3_r214 : last svn revision with default SSE flags (gcc compiler with CMake).
Aften_0.05 : version 0.05 with default settings for the compiler (gcc).
Aften_k8_r214 : last svn revision with extra flags (-march=k8 -O3 -pipe -fomit-frame-pointer)
time value in seconds

[DarkAvenger]

Thanks! In fact -O3 and -fomit-frame-pointer (OK this flag set by use in project file) are already set by cmake (though if you override CFLAGS you have to set them again, at least -O3 and -DNDEBUG.) So in the end the march setting gives you a small speed-up. But it's nice to see that svn version is again much faster than last release.

Oh another note: A 64bit compile is about 20% faster than a 32bit one on my x86_64 Linux system.

BTW, you can use "make VERBOSE=1" to see what cmake passes to gcc.

[tebasuna51]

Hi,
I just want to mention that my attempt at DRC encoding has been committed to Aften SVN. If anyone wants to try it out, the commandline option is -dynrng. Below is a snippet from the usage text.

[-dynrng #] Dynamic Range Compression profile
0 = Film Standard
1 = Film Light
2 = Music Standard
3 = Music Light
4 = Speech
5 = None (default)

-Justin
WARNING, also in your Aftenblog:
example using Film Standard:
aften -dnorm 27 -dynrng 0 test.wav test.ac3

example using Music Light:
aften -dnorm 17 -dynrng 3 test.wav test.ac3
But with aften -h we have
[-dynrng #] Dynamic Range Compression profile
0 = Film Light
1 = Film Standard
2 = Music Light
3 = Music Standard
4 = Speech
5 = None (default)
And I think this last is the correct way.
With this last parameters I make my test (http://forum.doom9.org/showthread.php?p=893111#post893111)

[danpos]

@Darkavenger

I'm having some difficulties in order to compile the last svn revision (214). I'm using Ubuntu 'Edgy' GNU/Linux with gcc 4.1.2 / yasm 0.4.0. This is what I got after do a 'checkout' on svn repository:


drwxr-xr-x 3 danpos danpos 4096 2006-10-31 21:54 aften
-rw-r--r-- 1 danpos danpos 2001 2006-10-31 21:53 bswap.h
-rw-r--r-- 1 danpos danpos 2615 2006-10-31 21:53 Changelog
-rw-r--r-- 1 danpos danpos 7611 2006-10-31 21:53 CMakeLists.txt
drwxr-xr-x 3 danpos danpos 4096 2006-10-31 21:54 CMakeModules
-rw-r--r-- 1 danpos danpos 3688 2006-10-31 21:53 common.h
-rwxr-xr-x 1 danpos danpos 18162 2006-10-31 21:53 configure
-rw-r--r-- 1 danpos danpos 26428 2006-10-31 21:53 COPYING
drwxr-xr-x 4 danpos danpos 4096 2006-10-31 21:54 libaften
-rw-r--r-- 1 danpos danpos 1833 2006-10-31 21:53 Makefile
-rw-r--r-- 1 danpos danpos 962 2006-10-31 21:53 README
drwxr-xr-x 3 danpos danpos 4096 2006-10-31 21:54 util

Now, how do I must to use the compilation commands in order to get an optimized version from my machine (Athlon XP 2600+)?

Thanks in advance,

[DarkAvenger]

See first post on this page.

Create your working dir and change into it and try:

CFLAGS="-march=athlon-xp -O3 -DNDEBUG" cmake ..
make

[danpos]

@DarkAvenger

Thanks for your reaction. As a matter of fact, I did read your first post out and seems that I'd not understood it correctly. I follow your advises and so got these results:


root@WOLF:/home/danpos/Pacotes/Aften# CFLAGS="-march=athlon-xp -O3 -DNDEBUG" cmake .
-- Check for working C compiler: gcc
-- Check for working C compiler: gcc -- works
-- Check size of void*
-- Check size of void* - done
Please do an out-of-tree build:
rm -f CMakeCache.txt; mkdir -p default; cd default; cmake ..; make
CMake Error: Error in cmake code at
/home/danpos/Pacotes/Aften/CMakeLists.txt:15:
MESSAGE in-tree-build detected
Current CMake stack: /home/danpos/Pacotes/Aften/CMakeLists.txt;/usr/share/CMake/Modules/CMakeCInformation.cmake
-- Configuring done


root@WOLF:/home/danpos/Pacotes/Aften# make
Makefile:4: config.mak: Arquivo ou diretório inexistente
touch config.mak
make -C libaften all
make[1]: Entrando no diretório `/home/danpos/Pacotes/Aften/libaften'
make[1]: Nada a ser feito para `all'.
make[1]: Saindo do diretório `/home/danpos/Pacotes/Aften/libaften'
make -C aften all
make[1]: Entrando no diretório `/home/danpos/Pacotes/Aften/aften'
cc -I. -I.. -I../libaften -I/libaften -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_ISOC9X_SOURCE -c -o aften.o aften.c
In file included from aften.c:25:
../common.h:105: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘log2tab’
../common.h:120: error: expected ‘)’ before ‘v’
In file included from ../common.h:129,
from aften.c:25:
../bswap.h:37: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘bswap_16’
../bswap.h:41: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘bswap_32’
../bswap.h:46: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘bswap_64’
In file included from aften.c:37:
wav.h:53: error: expected specifier-qualifier-list before ‘uint32_t’
aften.c: In function ‘main’:
aften.c:432: error: ‘uint8_t’ undeclared (first use in this function)
aften.c:432: error: (Each undeclared identifier is reported only once
aften.c:432: error: for each function it appears in.)
aften.c:432: error: ‘frame’ undeclared (first use in this function)
aften.c:440: error: ‘uint32_t’ undeclared (first use in this function)
aften.c:440: error: expected ‘;’ before ‘samplecount’
aften.c:496: error: ‘WavFile’ has no member named ‘channels’
aften.c:497: error: ‘WavFile’ has no member named ‘channels’
aften.c:497: error: ‘WavFile’ has no member named ‘ch_mask’
aften.c:498: error: ‘WavFile’ has no member named ‘sample_rate’
aften.c:503: error: ‘WavFile’ has no member named ‘read_format’
aften.c:514: error: ‘WavFile’ has no member named ‘channels’
aften.c:520: error: ‘samplecount’ undeclared (first use in this function)
aften.c:520: error: ‘bytecount’ undeclared (first use in this function)
aften.c:520: error: ‘t0’ undeclared (first use in this function)
aften.c:520: error: ‘t1’ undeclared (first use in this function)
aften.c:520: error: ‘percent’ undeclared (first use in this function)
aften.c:525: error: ‘WavFile’ has no member named ‘channels’
aften.c:538: error: ‘WavFile’ has no member named ‘sample_rate’
aften.c:539: error: ‘WavFile’ has no member named ‘samples’
aften.c:541: error: ‘WavFile’ has no member named ‘sample_rate’
aften.c:546: error: ‘WavFile’ has no member named ‘samples’
aften.c:547: error: ‘WavFile’ has no member named ‘samples’
aften.c:569: error: ‘WavFile’ has no member named ‘sample_rate’
make[1]: ** [aften.o] Erro 1
make[1]: Saindo do diretório `/home/danpos/Pacotes/Aften/aften'
make: ** [progs] Erro 2

Any ideas?

Thanks in advance,

[DarkAvenger]

Why don't you *read*? YOu need to create a work dir!

[danpos]

Why don't you *read*? YOu need to create a work dir!

Thanks, I think that now I did *read* it out correctly. :)

Regards,

[sdsumike619]

I just discovered this encoder from someone else. Can someone tell me what settings to use, boxes to check, etc. in order for me to be able to encode a regular wav file (no surround 5.1 or anything like that). I also want to encode the AC3 so that it has the same volume throughout as the original wav did.

It seems everytime I encode an AC3, no matter what software I'm using, the volume is always much lower. So someone on a another forum told me to use Aften and it could accomplish what I needed.

Thanks for the input

[Kurtnoise]

Can someone tell me what settings to use, boxes to check, etc. in order for me to be able to encode a regular wav file (no surround 5.1 or anything like that). I also want to encode the AC3 so that it has the same volume throughout as the original wav did.
Just use default settings...


btw, does anybody is interested by an update of the GUI ? To test mainly the dynamic range compression profiles...

[raquete]

btw, does anybody is interested by an update of the GUI ? To test mainly the dynamic range compression profiles...
i am Kurt,just tell me what to do and what you really want to know about the results.
i have lots of sources for test.
please update the GUI,is my favorite tool! :)

[LigH]

Of course we are interested. For example, in the german doom9/Gleitz board, a thread about recommendable settings was just started by katjarella, and Archimedes1 seems to know a good amount of background.

http://forum.gleitz.info/showthread.php?t=31827

[Kurtnoise]

Eine englische Übersetzung bitte...xD

So, AftenGUI 1.3 is available (http://kurtnoise.free.fr/index.php?dir=Aften/&file=AftenGUI-1.3.zip)...

changes : added DRC Profiles & Bits Allocation options.

[raquete]

Big thanks kurt.

[jruggle]

I can't access sources or others revisions, http://jbr.homelinux.org/aften/ don't work for me and http://sourceforge.net/projects/aften is outdated.

Hi all,
Sorry the daily builds are not working. My internet connection went down last week and due to financial constraints it may be another week or so before I'm back online. I'm posting now from the local library. I'll check back periodically to answer any questions that may come up.
-Justin

[DarkAvenger]

I am currently parallelizing aften. :) At least after hours of struggling I got the main locking to work. (I "never" used pthreads before and linux scheduling regarding threads is quite unfair.) While there is still some locking for some critical regions missing (ie output is distorted, but I know where it is missing), it seems to work now.

So on my dual core, encoding time for a test sample went down from 8sec to 4.9sec, which isn't too bad. It will probably be a bit slower once everything is correctly locked, but there is some room for optimization.

Are here any Linux users with dual core systems? Once I got it working (perhaps this week-end; I have a full time job now, so less time for hobbies) I'd like to have some testers. Just send me a pm with email addy and I'll contact you once I got something ready.

Oh, and does anybody know (in a portable way) how to find out at run-time, how many physical cores are available?

Of course I'll try to make it portable/win32 compatible. ;)

[Edit]
So, I managed to get binary identical output from the threaded version. So, I'll see whether I get it working in win32 and clean up the mess... Furthermore it seems I found some sort of bug in the bitalloc.

I needed to revert http://svn.sourceforge.net/viewvc/aften/libaften/bitalloc.c?r1=77&r2=78 to get identical results. (Un)Applying that patch gives me different results even with single-threaded version. So is the search faulty or is the "better start estimate" faulty? Justin?

[Edit]
OK, I got the Win32 Version going as well. But I noticed my gcc3.4 for mingw miscompiles it. Using msvc works and mingw-crosscompiler for linux based on gcc 4.1.2 works as well. I found gcc 4.1.1 (http://gda.utp.edu.co/~ceniza/GCC-4/4.1.1/) binary for mingw, which works as well. So time to ditch gcc 3.4.... As I don't have Windows installed on my dual core, I couldn't test whether the threading actualy works as promised. Using Wine under Linux it seems it does, though not as effective as the pthread version. (It seems windoes doesn't provide an equivalent to the pthread condition variable, so I used events. Well, Windows Vista seems to support the cond var, so converting pthread apps to vista seems to be straight forward...)

I found some means to detect number of available cpus, which I want to integrate, then I'll release some bins and an (unclean) patch.

[Edit]
Here it is: http://mitglied.lycos.de/darkav/download/aften_parallel.zip
You need to visit darkav.de.vu first and paste above url, otherwise you can't access the file.

[Kurtnoise]

*bump*...Does anybody have tested the parallel build from DarkAvenger ? Download mirror here (http://kurtnoise.free.fr/Aften/)...

btw, compilation failed with rev 215 :
[ 66%] Building C object CMakeFiles/aften_static.dir/libaften/x86/x86_sse3_mdct_dummy.obj
In file included from c:/temp/aften/libaften/x86/x86_sse3_mdct_dummy.c:2:
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c: In function `mdct_bitreverse':
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:430: warning: implicit declaration of function `_mm_castsi128_ps'
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:430: error: incompatible types in assignment
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:431: error: incompatible types in assignment
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:432: error: incompatible types in assignment
c:/temp/aften/libaften/x86/x86_sse_mdct_common.c:433: error: incompatible types in assignment
make[2]: *** [CMakeFiles/aften_static.dir/libaften/x86/x86_sse3_mdct_dummy.obj] Error 1
make[1]: *** [CMakeFiles/aften_static.dir/all] Error 2
make: *** [all] Error 2


edit: mmh...compilation is fine with gcc 4.1.1 but not with 3.4.5.

[DarkAvenger]

gcc 3.4.5 doesn't provide the _mm_castsi128_ps intrinsic, but cmake should detect it and activate the work-around. I don't get why it breaks now. Could you mail/post the generated config.h? Another thought: Is there some other config.h around (eg in aften root)? This could be a trouble-maker, as well. I added the binary dir first in the CMake project file now, which should fix this.

BTW, cmake 2.4.4 was released. I haven't tested yet, but perhaps this version doesn't segfault anymore, so cvs cmake wouldn't be needed on windows?

[Kurtnoise]

Another thought: Is there some other config.h around (eg in aften root)? This could be a trouble-maker, as well.
aaa yes...That was the problem. Thanks. It works fine now.

BTW, cmake 2.4.4 was released. I haven't tested yet, but perhaps this version doesn't segfault anymore, so cvs cmake wouldn't be needed on windows?
I used to use always the last cvs checkout. So, no problem with this.

[raquete]

@ Kurt
please,can you give me a deep explanation to adjust the Misc tab,the Filters & Overrides Tab in AftenGUI1.3?
and what After version i use with this GUI?
i really need and apriciate your help.

strange comparison with Creedence - Pagan Baby:
-31dB adjusted / DRC none...same source(.wav)
vegas: AC3 start too loud(crispy) and descrease few seconds later
sfsencode: AC3 start loud,decrease few seconds and when the voice start decrease the volume a little more.
AftenGUI1.1: volume start normal and stay normal in the whole music just like the source!
in the end....trashing vegas and sencode again and calling for help to adjust the tabs,like i'm doing now.

best regards!
:)

[Skelsgard]

Could you post a 30 secs clip of the original WAV for that Credence song?

Cheers

[raquete]

of course.
do you want with fade out or abrupt cut?

[Skelsgard]

Abrupt cut will do just fine. I´m more interested in the section where you find the volume inconsistencies between encoders.

Cheers.

[raquete]

here it is Skelsgard: http://rapidshare.com/files/5007399/raq01.rar.html

the issue happen in the begining of the music where the singer start to sing with softencode and vegas.
i reinstall softencode and using this little sample the issue happen again.
vegas is trashed forever...who needs of 172Mb installed to get audio issues? lol

with AftenGUI1.1 using aften.exe from 7/30/2006 the resulting sound is perfect!!!
thanks for your interest! :)

@ Kurt
my friend,without guide i got cool results,now imagine what i can get if you answer my requests of the adjusts in AftenGUI1.3
:thanks:

[Skelsgard]

I´ve tested the sample and it was impossible for me to find these inconsistencies between encoders using the same parameters.
I´m posting the options used for each program, and in the next post the results (AC3s and decoded WAVs).

Aften 0.05 - Rev. 214
http://img168.imageshack.us/img168/8181/aften01qp7.jpghttp://img490.imageshack.us/img490/7334/aften02cp0.jpg
http://img212.imageshack.us/img212/9213/aften03eg6.jpghttp://img177.imageshack.us/img177/5536/aften04gh6.jpg

Soft Encode
http://img216.imageshack.us/img216/9730/softencode01pp3.jpg
http://img216.imageshack.us/img216/737/softencode02mt7.jpg

Vegas 7
http://img175.imageshack.us/img175/4807/vegas01eg2.jpghttp://img83.imageshack.us/img83/6606/vegas02jg2.jpg
http://img175.imageshack.us/img175/1368/vegas03kf6.jpghttp://img175.imageshack.us/img175/4472/vegas04zu7.jpg

Cheers.

[Skelsgard]

These are the AC3 files obtained:
Aften AC3 http://www.badongo.com/file/1783659
Soft Encode AC3 http://www.badongo.com/file/1783694
Vegas AC3 http://www.badongo.com/file/1783695

These are AC3-decoded WAVs:
Aften http://www.badongo.com/file/1783753
Soft Encode http://www.badongo.com/file/1783791
Vegas http://www.badongo.com/file/1783862


These are the peak files generated from the AC3-decoded WAVs, in order from top to bottom:
- Original
- Aften encoded
- Soft Encoded encoded
- Vegas 7 Encoded
http://img85.imageshack.us/img85/53/peakspu6.jpg

The WAVs were decoded with Soft Encode thru Open --> Dolby Digital (decode to .PCM).
No apparent difference between the files can be noticed at first sight in the peak images. More subtle differences might be found due to the inherent caracteristics of the encoders.

The difeerences you find strike me as odd. You may have been using DRC in the Vegas and Soft Encode samples, raquete.
Could you test again with the options given in the pics?

Cheers.

Edited:
I´ve just realized that the Vegas sample was encoded as 48kHz. This are the 44.1 kHz samples.
AC3 http://www.badongo.net/file/1784039
Decoded http://www.badongo.net/file/1784067

[raquete]

wow..don't need it all.

differences and problems:
you used aftengui1.3,i used aftengui1.1(re read my posts)
you adjusted in the gui to 192Kbps, i to 448Kbps
you encoded AC3-2.0,i 5.1!
dial.norn -31 are equal mine
Aftengui1.1 don't have adjusts like 1.3 version,then...

as you encoded 2.0 lots of adjusts in softencode are differents.
how was used:
448Kbps
48K
3/2 lfe unchecked
complete main
-31dB
20.30KHz
-3dB center mix
-3dB surround mix
copyright bit
original bit stream
info exists
mix level 95 dB SPL
small room flat monitor
DC highpass filter
Bandwidth low-pass filter
You may have been using DRCno DRC(none)! we are changing informations,see what i posted-31dB adjusted / DRC none...i was clever.

The WAVs were decoded with Soft Encode...as i'm scared about softencode i don't will use it to decode.
Could you test again with the options given in the
pics?no,is needed to follow the adjusts that i posted to reproduce the issue and,as you read,don't have anything wrong in the adjusts that i used and no more vegas,was trashed.install 172Mb again to do a single test is too much for my beauty. lol
i will host the results.
softencode and aftengui1.1 only ok?

big thanks for your tests but we have to restart :p

edit: typos

[tebasuna51]

@raquete, I make the test with your sample but using Aften, SoftEncode and Scenarist (I haven't Vegas).

Like Skelsgard I don't found significative differences between the three encodes. I use the same parameters than Skelsgard and don't understand your differences:

- The sample is stereo then must be encoded like 2.0, with 192 Kb/s (default for 2.0)

- The sample is 44.1 KHz and must be encoded without change, in Aften you can't modify the samplerate, but in SoftEncode if you modify the samplerate to 48 KHz SoftEncode assume the original samplerate is 48 KHz (don't make any translation, check the duration).

- Other SoftEncode parameters must be the same than Aften then:
Don't use DC highpass filter
Don't use Bandwidth low-pass filter
And DRC(none) is the same than DRC none... (BTW using or not DRC in encode phase is the same if DRC is not used in the decode phase)

- At last I used Azid (without DRC) to decode the three ac3. SoftEncode can be used if you select in decoder options:
Dynamic Range Compression: Custom mode, analog dialog norm.

I think the problem can be the 44.1 KHz <-> 48 KHz

[raquete]

dear friend tebasuna,
is hard to answer what you can imagine how i did.
i was clear in what i posted and how was done.
the source is stereo because Skelsgard was asking for the source (read his post how he needed the source) but i extract channels and did ac3 5.1 in 48K.
as i can't explain in clever words, better is screenshots and my samples from sources(LR,CLFE,Ss)waves and the ac3 results from aften and from softencode.
then you will see the adjusts used in softencode and hear clearly the issue!
just give me few minutes. :)

[raquete]

ok,here they are:
sources (LR,CLFE,SLSR) : http://rapidshare.com/files/5144213/sources.rar

AC3 from AftenGUI1.1 : http://rapidshare.com/files/5142068/aftengui.rar

AC3 from softencode : http://rapidshare.com/files/5142468/softencode.rar

now hear what happens in 22 seconds when the singer start to sing in the AC3 files,compare with the source and tell me what is equal,what is different.

adjusts used in softencode.
the manual recommend to use DC highpass filter and Bandwidth low-pass filter,then i can't see reason to uncheck this options

http://img292.imageshack.us/img292/6126/01softencodesh9.png

http://img450.imageshack.us/img450/4337/02softencodefo7.png

now please,use the sources posted,adjust softencode like the pictures and compare the results.

[Skelsgard]

You came out somewhat agressive on that last post to me.
This is what you wrote:
strange comparison with Creedence - Pagan Baby:
-31dB adjusted / DRC none...same source(.wav)
vegas: AC3 start too loud(crispy) and descrease few seconds later
sfsencode: AC3 start loud,decrease few seconds and when the voice start decrease the volume a little more.
AftenGUI1.1: volume start normal and stay normal in the whole music just like the source!
in the end....trashing vegas and sencode again and calling for help to adjust the tabs,like i'm doing now.

In no place you state that you´ve upmixed it to 5.1 or any other parameters other than -31dB and DRC none nor that you´ve encoded using previously stated options, so I assumed that you´ve tested all programs with the same parameters, respecting the caracteristics of the audio (channel number, bandwidth, etc).

So, using the parameters from the pics given by raquete...
these are the AC3 samples obtained:
http://www.badongo.com/file/1786841
http://www.badongo.com/file/1786842
http://www.badongo.com/file/1786843

a peak image of all four WAVs (this time decoded with a non-biased decoder: BeHappy) in order from left to right:
raquete´s sample - Aften - Soft Encode - Vegas
http://img244.imageshack.us/img244/1479/peaks51ar7.jpg

Again, no visual or audible differences.

Cheers.

[raquete]

You came out somewhat agressive on that last post to me.never..please excuse is my poor english or some differents regionals language issues.
excuse please,i still don't read the remainder of your post but was afraid of what you are feeling.
excuse me

[raquete]

i re read my old post and don't read nothing agressive against you,see that was not edited!

about your screenshots: i see differences(phases,levels...),even the little lfe show clear differences.

your ac3 samples: the ac3 softencode sample have the same issue in 22 seconds where the singer start.
the sound compress/normalize(fast decay in the whole music in this moment),exactly like in my sample.
another detail is that you adjusted softencode to 105dB SPL(in my screenshot show 95)
i play the samples in mpclassic and in neroshowtime!!!

someone could download and hear the samples,please?
tebasuna?

[tebasuna51]

I know you aren't agressive, maybe a little vehement in yours comments and admit, in your first post are not clear the test parameters.

After that, yes, I download the samples and like Skelsgard I don't found visual differences but seems there are clear audible differences at singer start (22 sec).

I say "seems" because I have attached a old audio equipment with only DPL I decoder and to play 5.1 audio I need:
5.1 -> ffdshow -> dpl I -> Audio Eq -> 6 speakers
To hear 5.1 I need burn a CD/DVD and go out my PC place. Maybe tomorrow.

Then seems there are real differences. And I reproduce the encodes with same issue. I tested with/without filters in SoftEncode. Also with Mid/Side versus Independent in AftenGui -> Misc -> Stereo ReMatrixing (I don't know if is used in 5.1, but in 2.0 make different result).

To be continued...

[raquete]

I know you aren't agressivethanks for recognize and for the friendship.

in your first post are not clear the test parameters.the first intention was to tell about the issues and second that was used the same parameters in the encoders...of course i forgot about channels,was not important in that moment.

Maybe tomorrow. ... To be continued...all right,waiting for you and for Skelsgard.

regards

[Skelsgard]

Is just a trivial misunderstanding.
About my tests, I´ll check again, might have missed something there.
If I can, I´ll try to check with Vegas 6 encoder, but not possible right now.

Cheers.

[Skelsgard]

I´m playing thru BSPlayer with AC3filter 1.09 and honestly can´t hear any difference between the samples as either 5.1 nor downmixed 2.0.
I´m gonna try to ABX them later.

Cheers.

[raquete]

hy.
I´m gonna try to ABX them later.my preference is your ears only.
later(another thread?) we talk about ABX ok? ;)

[raquete]

from Skerlsgard's AC3 samples... http://forum.doom9.org/showthread.php?p=907623#post907623

open one by one in differents pages:
aften: http://img82.imageshack.us/my.php?image=credenceaften51hq4.png
softencode: http://img90.imageshack.us/my.php?image=credencese51hi3.png
vegas: http://img90.imageshack.us/my.php?image=credencevegas51fs9.png

the same screenshots from:
aften:
http://img82.imageshack.us/img82/6075/credenceaften51hq4.png

softencode:
http://img90.imageshack.us/img90/628/credencese51hi3.png

vegas:
http://img90.imageshack.us/img90/8144/credencevegas51fs9.png

each vertical line mark 10 seconds,see how change after 22 seconds,in se and vegas the sounds seems compressed.
from aften is like the source:where the sound is low remains low,where encrease remains encreasing and where is loud remains loud.
i clearly see differences...(listenig too)

i need comments and i'm still waiting one little "how to" to use adjust the AftenGUI1.3 and what means each adjust in the tabs...please!

thanks so much
;)

[Skelsgard]

I´ve re-tested the samples and I absolutely see the difference you´re talking about. I can´t understand how I´ve missed it. Probably the short display of the peak files led me to confussion.
I still have trouble finding audible differences between Aften and Vegas AC3 samples, even in the Soft Encode sample too. At least, in the "compressed" way you talk about.

Cheers.

[DarkAvenger]

I uploaded an updated patch of parallelized aften (against rev 216):

http://mitglied.lycos.de/darkav/download/aften_parallel2.zip

(again, visit darkav.de.vu first. I was too lazy to include binaries this time. ;) Please don't compile with mingw gcc3.4, use gcc 4.x.)

- buildable without threads
- if one cpu (core) is detected, unthreaded version is used
- mdct uses one global context and per thread buffers, instead of duplicating tables to every thread

I really hope Justin gets his internet connection back, so that he can review it and I can start merging this stuff...

[raquete]

impossible to download aften_parallel2.zip.
something wrong in lycos.de (message from IE)

[DarkAvenger]

See what's written below the link. You must paste the address after vising the home. It is an anti-leech protection.

You also read that no binaries are included, ie you have to compile yourself?

[DarkAvenger]

Probably the last functional update (again, patch only):

http://mitglied.lycos.de/darkav/download/aften_parallel3.zip

- parallelized filter stage using more locks

In fact now aften should now be completely parallelized. Only more clean-ups for merging are left.