Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Flake vs Flac and the -Ax2 switch (Read 29393 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Flake vs Flac and the -Ax2 switch

With reference to Synthetic Soul's Lossless comparison I noticed flac -8 -Ax2 compresses better and has faster decompression than -8. I batched half a dozen albums with both settings and the files size are identical. Tried on both Linux and Win with the same results. Anyone had any experience with -Ax2?

I've also been trying out Flake. With the better speeds (and noticeable compression using the higher switches) than the reference encoder, should this encoder be used, or is it unreliable. I notice it doesn't feature a verify option, but apart from that, is there something else about Flake for it not to be recommended.

Thanks.

Flake vs Flac and the -Ax2 switch

Reply #1 –
It depends on the kind of music whether a certain kind of apodization filter compresses better. Usually the default settings are pretty good. As far as I know, the Flake encoder has currently no problems and is quite error robust.
Music: sounds arranged such that they construct feelings.

Flake vs Flac and the -Ax2 switch

Reply #2 –
Quote from: ManekiNeko on
I batched half a dozen albums with both settings and the files size are identical.

Files size are almost identical. Compression gain is very small: 0.11%.
For example,
FLAC 8 -Ax2 - 400 MB
FLAC 8 - 400.44 MB

Generally I use -8 -Ax2 not only because of +0.11% compression gain but because decoding speed is slightly higher which is useful for future lossy trancode for player. And I have a little collection of CDs and maybe one day I have to rip one only 1 CD to FLAC. There is no difference 2 vs 3 mins if coffee time is about 5 mins.

Flake vs Flac and the -Ax2 switch

Reply #3 –
Quote from: IgorC on
Quote from: ManekiNeko on
I batched half a dozen albums with both settings and the files size are identical.

Files size are almost identical. Compression gain is very small: 0.11%.
For example,
FLAC 8 -Ax2 - 400 MB
FLAC 8 - 400.44 MB


OK. I've had some more time to do some more tests. I encoded 3 more albums from wav to flac, with and without -Ax2. I'm still getting identical results. I purposely chose some quiet albums to highlight better compression.

Flac -8 Log

Code: [Select]
flac -8 -Ax2 *.wav

flac 1.2.1, Copyright (C) 2000,2001,2002,2003,2004,2005,2006,2007  Josh Coalson
flac comes with ABSOLUTELY NO WARRANTY.  This is free software, and you are
welcome to redistribute it under certain conditions.  Type `flac' for details.

01 Ophelia.wav: wrote 24636257 bytes, ratio=0.451
02 Life Is Sweet.wav: wrote 33682625 bytes, ratio=0.615
03 Kind & Generous.wav: wrote 23922555 bytes, ratio=0.548
04 Frozen Charlotte.wav: wrote 26079007 bytes, ratio=0.458
05 My Skin.wav: wrote 30572115 bytes, ratio=0.525
06 Break Your Heart.wav: wrote 28175170 bytes, ratio=0.556
07 King of May.wav: wrote 25433586 bytes, ratio=0.578
08 Thick as Thieves.wav: wrote 41283485 bytes, ratio=0.562
09 Effigy.wav: wrote 11950741 bytes, ratio=0.451
10 The Living.wav: wrote 16735856 bytes, ratio=0.483
11 When They Ring the Golden Bells.wav: wrote 49785363 bytes, ratio=0.492


And the flac -8 log:

Code: [Select]
flac -8 *.wav

flac 1.2.1, Copyright (C) 2000,2001,2002,2003,2004,2005,2006,2007  Josh Coalson
flac comes with ABSOLUTELY NO WARRANTY.  This is free software, and you are
welcome to redistribute it under certain conditions.  Type `flac' for details.

01 Ophelia.wav: wrote 24636257 bytes, ratio=0.451
02 Life Is Sweet.wav: wrote 33682625 bytes, ratio=0.615
03 Kind & Generous.wav: wrote 23922555 bytes, ratio=0.548
04 Frozen Charlotte.wav: wrote 26079007 bytes, ratio=0.458
05 My Skin.wav: wrote 30572115 bytes, ratio=0.525
06 Break Your Heart.wav: wrote 28175170 bytes, ratio=0.556
07 King of May.wav: wrote 25433586 bytes, ratio=0.578
08 Thick as Thieves.wav: wrote 41283485 bytes, ratio=0.562
09 Effigy.wav: wrote 11950741 bytes, ratio=0.451
10 The Living.wav: wrote 16735856 bytes, ratio=0.483
11 When They Ring the Golden Bells.wav: wrote 49785363 bytes, ratio=0.492


Finally, I ran Flake -8. According to the Flake Website it should compress higher. But my results show the opposite:

Code: [Select]
flake -8 *.wav

Flake: FLAC audio encoder
version 0.11
(c) 2006-2007 Justin Ruggles

block time: 105ms
variable block size: none
prediction type: levinson-durbin
prediction order: 1,12
partition order: 0,8
order method: 4-level
stereo method: mid-side
header padding: 4096

input file:  "01 Ophelia.wav"
output file: "01 Ophelia.flac"
Signed 16-bit 44100 Hz stereo
samples: 13651008 (5m9.546s)
block size: 4608
progress: 100% | ratio: 0.452 | bitrate: 637.4 kbps | bytes: 24636971 

input file:  "02 Life Is Sweet.wav"
output file: "02 Life Is Sweet.flac"
Signed 16-bit 44100 Hz stereo
samples: 13690404 (5m10.440s)
block size: 4608
progress: 100% | ratio: 0.617 | bitrate: 870.7 kbps | bytes: 33753494 

input file:  "03 Kind & Generous.wav"
output file: "03 Kind & Generous.flac"
Signed 16-bit 44100 Hz stereo
samples: 10916220 (4m7.533s)
block size: 4608
progress: 100% | ratio: 0.550 | bitrate: 776.8 kbps | bytes: 24006196 

input file:  "04 Frozen Charlotte.wav"
output file: "04 Frozen Charlotte.flac"
Signed 16-bit 44100 Hz stereo
samples: 14242536 (5m22.960s)
block size: 4608
progress: 100% | ratio: 0.458 | bitrate: 646.1 kbps | bytes: 26047212 

input file:  "05 My Skin.wav"
output file: "05 My Skin.flac"
Signed 16-bit 44100 Hz stereo
samples: 14560644 (5m30.173s)
block size: 4608
progress: 100% | ratio: 0.526 | bitrate: 742.8 kbps | bytes: 30649617 

input file:  "06 Break Your Heart.wav"
output file: "06 Break Your Heart.flac"
Signed 16-bit 44100 Hz stereo
samples: 12670224 (4m47.306s)
block size: 4608
progress: 100% | ratio: 0.559 | bitrate: 788.8 kbps | bytes: 28306993 

input file:  "07 King of May.wav"
output file: "07 King of May.flac"
Signed 16-bit 44100 Hz stereo
samples: 10994424 (4m9.306s)
block size: 4608
progress: 100% | ratio: 0.579 | bitrate: 817.8 kbps | bytes: 25472016 

input file:  "08 Thick as Thieves.wav"
output file: "08 Thick as Thieves.flac"
Signed 16-bit 44100 Hz stereo
samples: 18359124 (6m56.306s)
block size: 4608
progress: 100% | ratio: 0.564 | bitrate: 795.3 kbps | bytes: 41372650 

input file:  "09 Effigy.wav"
output file: "09 Effigy.flac"
Signed 16-bit 44100 Hz stereo
samples: 6622644 (2m30.173s)
block size: 4608
progress: 100% | ratio: 0.448 | bitrate: 632.1 kbps | bytes: 11861876 

input file:  "10 The Living.wav"
output file: "10 The Living.flac"
Signed 16-bit 44100 Hz stereo
samples: 8668884 (3m16.573s)
block size: 4608
progress: 100% | ratio: 0.483 | bitrate: 682.2 kbps | bytes: 16740442 

input file:  "11 When They Ring the Golden Bells.wav"
output file: "11 When They Ring the Golden Bells.flac"
Signed 16-bit 44100 Hz stereo
samples: 25276356 (9m33.160s)
block size: 4608
progress: 100% | ratio: 0.494 | bitrate: 697.0 kbps | bytes: 49924335


Summary. flac -8 and flac-8 -Ax2 are identical size and flake -8 is larger (but should be smaller). I have repeated this several times with different albums with the same result. Can somebody else run a similar test and produce different results?

On a side note, is flake -8 the highest setting which is compliant with the flac reference encoder.

Thanks guys.

Flake vs Flac and the -Ax2 switch

Reply #4 –
It can't be 100.00% identical. 
Try '-8 -A tukey(0.5) -A flattop'  instead of '-8 -Ax2'

Flake vs Flac and the -Ax2 switch

Reply #5 –
Quote from: ManekiNeko on
On a side note, is flake -8 the highest setting which is compliant with the flac reference encoder.

All settings are "compliant".  Levels 0 to 8 use approx. the same encoding settings as the FLAC reference encoder.  When keeping the same parameters there is not much that can be done to increase compression.  The main improvement in those settings is speed.  The higher settings focus more on compression.  9 is the one I use most often since it generates Subset-compliant files and is still pretty fast.  Also, the latest SVN version might do a little better than version 0.11 in terms of compression.  In version 0.11, try: flake -9 -b 4096 -v 1

Flake vs Flac and the -Ax2 switch

Reply #6 –
@ IgorC

Thanks, yes that worked. I had to leave out the brackets on Fedora 10, eg. -8 -A tukey 0.5 -A flattop to make it work. I take it that this command is the same as -Ax2? Strange that the -Ax2 switch does nothing on Linux (tested with Flac 1.21 on Fedora 10 and Ubuntu 9.04).

edit I didn't read the logs carefully:

flac -8 -A tukey 0.5 -A flattop *.wav: Encodes fine but receive 'ERROR: can't open input file 0.5: No such file or directory'
flac -8 -A tukey(0.5) -A flattop 01*.wav: Doesn't encode. Receive bash: syntax error near unexpected token `('
flac -8 -A tukey -A flattop 01*.wav: Encodes with no errors just missing the (0.5) option

@ Justin

Thank you. Using flake 0.11 with flake -9 -b 4096 -v 1 produced nice compression at very fast speeds. I do get the following warning:

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
WARNING! The chosen encoding options are
not FLAC Subset compliant. Therefore, the
encoded file(s) may not work properly with
some FLAC players and decoders.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

I guess that is standard message and nothing to be concerned about for pc use. I also ran a straight 'flake -9' and the files sizes are still bigger than 'flac -8', but of course encoding time was a lot quicker.

I would like to test out the Flake SVN but can't see a 'make' file. Do you have a static linux x86 binary I could try?

Thanks

Flake vs Flac and the -Ax2 switch

Reply #7 –
Quote from: ManekiNeko on
Summary. flac -8 and flac-8 -Ax2 are identical size

Quote from: IgorC on
It can't be 100.00% identical. 
Try '-8 -A tukey(0.5) -A flattop'  instead of '-8 -Ax2'

I've just skimmed this thread, so ignore me if I'm stating something that's previously been raised.

AFAIK "-Ax2" is shorthand to state "two -A switches were used".  I suspect the reason that -8 and -8 -Ax2 are identical is that FLAC is ignoring the invalid -Ax2 switch.

If you navigate to the -Ax2 settings page in my comparison you will note that the actual switches used were -8 -A tukey(0.5) -A flattop (a combination that seemed to work well in my tests).

Apologies for any confusion caused by my notation.

I'm on a horse.

Flake vs Flac and the -Ax2 switch

Reply #8 –
Quote from: ManekiNeko on
@ Justin

Thank you. Using flake 0.11 with flake -9 -b 4096 -v 1 produced nice compression at very fast speeds. I do get the following warning:

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
WARNING! The chosen encoding options are
not FLAC Subset compliant. Therefore, the
encoded file(s) may not work properly with
some FLAC players and decoders.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

I guess that is standard message and nothing to be concerned about for pc use. I also ran a straight 'flake -9' and the files sizes are still bigger than 'flac -8', but of course encoding time was a lot quicker.

For PC use, no you shouldn't have to worry about it.  I think that at the time of version 0.11 I was under the impression that variable block size was not Subset compliant.  But there is nothing in the format description which says this, so I'm pretty sure it's ok for streaming as well.

Quote
I would like to test out the Flake SVN but can't see a 'make' file. Do you have a static linux x86 binary I could try?

Flake uses CMake.  I can make a linux x86_64 static binary later today, but probably not x86.

Flake vs Flac and the -Ax2 switch

Reply #9 –
Thanks for offering to build the latest svn. Although I can't use the 64bit binary, I'm sure many here will be interested.

I built the svn using cmake. Although encoding works, and does indeed compression better with standard -9 option, I am getting "Error writing frame to output throughout 1 to 100%.

I used:

Code: [Select]
$ mkdir build
$ cd build
$ cmake ..
$ make


which produces some warnings but completes. (i've PM'ed you the log)

Flake vs Flac and the -Ax2 switch

Reply #11 –
Quote from: ManekiNeko on
Thanks for offering to build the latest svn. Although I can't use the 64bit binary, I'm sure many here will be interested.

I built the svn using cmake. Although encoding works, and does indeed compression better with standard -9 option, I am getting "Error writing frame to output throughout 1 to 100%.

Yeah, that is a known issue that I haven't examined thoroughly yet.  The output should still be ok though.  I'll try to take some time to look into it soon.

Flake vs Flac and the -Ax2 switch

Reply #12 –
I've noticed that on my Athlon XP 1800+ system, FLAC 1.2.1 compresses faster and better than FLAKE.

For example, The Doors "The End", full version (11 minutes, 42 seconds) yields:

Uncompressed WAV = 123,856,364 bytes

FLAKE v 0.11
==========
Options: -6

(implies -b 4608 and -r 8)
Compress time: 25 seconds
FLAC file size:  65,118,997 bytes
Ratio: 0.526


FLAC 1.2.1
==========
Options: -6

Compress time: 20 seconds
FLAC file size: 64,632,552 bytes
Ratio: 0.522


FLAC 1.2.1
==========
Options: -6 -b 4608 -r 8

(matches FLAKE's -6 as much as possible)

Compress time: 22 seconds
FLAC file size: 64,623,200 bytes
Ratio: 0.522



I think I'll stick with FLAC 1.2.1, since FLAKE right now just takes longer to produce larger files on my system(s).

Flake vs Flac and the -Ax2 switch

Reply #13 –
Quote from: WonderSlug on
I think I'll stick with FLAC 1.2.1, since FLAKE right now just takes longer to produce larger files on my system(s).


Yes, I'm getting similar ratios with flake 0.11 but the latest SVN is producing smaller files. I hope Justin can figure out the encode error messages.

Flake vs Flac and the -Ax2 switch

Reply #14 –
Latest SVN version? Can you give a link to binaries?

Flake vs Flac and the -Ax2 switch

Reply #16 –
Justin has just updated flake to SVN -r244 and it compiles and runs perfectly. Thanks Justin

Flake vs Flac and the -Ax2 switch

Reply #17 –
Quote from: ManekiNeko on
Justin has just updated flake to SVN -r244 and it compiles and runs perfectly. Thanks Justin

No problem.  I also added the long-overdue support for removing unused low-order bits, which should allow Flake to work with lossyWAV (I haven't tested that yet though).

Flake vs Flac and the -Ax2 switch

Reply #18 –
Quote from: ManekiNeko on
[...]
edit I didn't read the logs carefully:

flac -8 -A tukey 0.5 -A flattop *.wav: Encodes fine but receive 'ERROR: can't open input file 0.5: No such file or directory'
flac -8 -A tukey(0.5) -A flattop 01*.wav: Doesn't encode. Receive bash: syntax error near unexpected token `('
flac -8 -A tukey -A flattop 01*.wav: Encodes with no errors just missing the (0.5) option


This is not a problem either with Linux or flac linux version, you are not following the bash rules. The characters ( and ) are being interpreted as especial bash characters. You can use the escape character \ to get rid of this problem. Read more here http://tldp.org/LDP/Bash-Beginners-Guide/h...sect_03_03.html

The correct command should be something like:
Code: [Select]
 
flac -8 -A tukey\(0.5\) -A flattop *.wav

 

Flake vs Flac and the -Ax2 switch

Reply #19 –
Quote
The correct command should be something like:
Code: [Select]
 
flac -8 -A tukey\(0.5\) -A flattop *.wav


Thanks PajaRo.

Yes, that was a little silly of me. I actually realized earlier today and used:

Code: [Select]
-8 -A tukey'(0.5)' -A flattop
which worked perfectly.

I also ran your suggestion:

Code: [Select]
flac -8 -A tukey\(0.5\) -A flattop *.wav
which is also perfect.

Both options returned identical file sizes.

Flake vs Flac and the -Ax2 switch

Reply #20 –
Just in case someone needs a precompiled build of Flake, I've made some unix builds from latest SVN r248:

OS X 10.3.9 (PPC)
OS X 10.4 (PPC)
OS X 10.4 (PPC64)
OS X 10.4 (i386)
OS X 10.4 (x86_64)
OS X 10.5 (PPC)
OS X 10.5 (PPC64)
OS X 10.5 (i386)
OS X 10.5 (x86_64)
FreeBSD 7.x (i386)
Linux (i386)

Maybe I'll add a few more later. Currently my Open Solaris is b!tching with me (probably needs a reinstall), and my Win32 is hating me (CMake ain't working with MinGW, Cygwin, OpenWatcom, MS Visual C++ 2005, MS Visual C++ 2008 and Borland C 5.5).
WavPack 4.50.1 -hhx6 | LAME 3.98.2 -V 0

Flake vs Flac and the -Ax2 switch

Reply #21 –
And here's an Open Solaris build:
Sun OS 5.11 (i386)
WavPack 4.50.1 -hhx6 | LAME 3.98.2 -V 0

Flake vs Flac and the -Ax2 switch

Reply #22 –
Ooops ... The PPC Mac OS X builds aren't working! 
I was noticing this when when trying to build on  PPC Linux, that there are a lot warnings which don't appear on the other OSes (even not on my PPC Mac):

Code: [Select]
root@freetibet:/home/frank/flake-enc/build# make                
Scanning dependencies of target pcm_io                          
[  5%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/aiff.o                 
[ 11%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/byteio.o               
[ 17%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/convert.o              
[ 23%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/formats.o              
[ 29%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/pcm_io.o               
[ 35%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/raw.o                  
[ 41%] Building C object CMakeFiles/pcm_io.dir/libpcm_io/wav.o                  
Linking C static library libpcm_io.a                                            
[ 41%] Built target pcm_io                                                      
Scanning dependencies of target flake_static                                    
[ 47%] Building C object CMakeFiles/flake_static.dir/libflake/crc.o             
[ 52%] Building C object CMakeFiles/flake_static.dir/libflake/encode.o          
/home/frank/flake-enc/libflake/encode.c: In Funktion »write_vorbis_comment«:    
/home/frank/flake-enc/libflake/bitio.h:90: Warnung: assuming pointer wraparound does not occur when comparing P +- C1 with P +- C2                              
/home/frank/flake-enc/libflake/bitio.h:90: Warnung: assuming pointer wraparound does not occur when comparing P +- C1 with P +- C2                              
/home/frank/flake-enc/libflake/encode.c: In Funktion »flake_encode_init«:       
/home/frank/flake-enc/libflake/bitio.h:90: Warnung: assuming pointer wraparound does not occur when comparing P +- C1 with P +- C2                              
/home/frank/flake-enc/libflake/bitio.h:90: Warnung: assuming pointer wraparound does not occur when comparing P +- C1 with P +- C2                              
[ 58%] Building C object CMakeFiles/flake_static.dir/libflake/lpc.o             
/home/frank/flake-enc/libflake/lpc.c: In Funktion »lpc_calc_coefs«:             
/home/frank/flake-enc/libflake/lpc.c:85: Warnung: assuming signed overflow does not occur when simplifying conditional to constant                              
[ 64%] Building C object CMakeFiles/flake_static.dir/libflake/md5.o             
[ 70%] Building C object CMakeFiles/flake_static.dir/libflake/metadata.o        
/home/frank/flake-enc/libflake/metadata.c: In Funktion »flake_write_streaminfo«:
/home/frank/flake-enc/libflake/bitio.h:90: Warnung: assuming pointer wraparound does not occur when comparing P +- C1 with P +- C2                              
[ 76%] Building C object CMakeFiles/flake_static.dir/libflake/optimize.o        
[ 82%] Building C object CMakeFiles/flake_static.dir/libflake/rice.o            
[ 88%] Building C object CMakeFiles/flake_static.dir/libflake/vbs.o             
Linking C static library libflake_static.a                                      
[ 88%] Built target flake_static                                                
Scanning dependencies of target flake_exe                                       
[ 94%] Building C object CMakeFiles/flake_exe.dir/flake/flake.o                 
/home/frank/flake-enc/flake/flake.c: In Funktion »parse_number«:                
/home/frank/flake-enc/flake/flake.c:124: Warnung: assuming signed overflow does not occur when simplifying conditional to constant                              
/home/frank/flake-enc/flake/flake.c: In Funktion »main«:                        
/home/frank/flake-enc/flake/flake.c:287: Warnung: assuming signed overflow does not occur when simplifying conditional to constant                              
/home/frank/flake-enc/flake/flake.c:302: Warnung: assuming signed overflow does not occur when simplifying conditional to constant                              
/home/frank/flake-enc/flake/flake.c:303: Warnung: assuming signed overflow does not occur when changing X +- C1 cmp C2 to X cmp C1 +- C2                        
/home/frank/flake-enc/flake/flake.c:303: Warnung: assuming signed overflow does not occur when changing X +- C1 cmp C2 to X cmp C1 +- C2                        
/home/frank/flake-enc/flake/flake.c:527: Warnung: assuming signed overflow does not occur when changing X +- C1 cmp C2 to X cmp C1 +- C2                        
/home/frank/flake-enc/flake/flake.c:539: Warnung: assuming signed overflow does not occur when changing X +- C1 cmp C2 to X cmp C1 +- C2                        
Linking C executable flake                                                      
[ 94%] Built target flake_exe                                                   
Scanning dependencies of target wavinfo                                         
[100%] Building C object CMakeFiles/wavinfo.dir/util/wavinfo.o                  
Linking C executable wavinfo                                                    
[100%] Built target wavinfo                                                     
root@freetibet:/home/frank/flake-enc/build# make install                        
[ 41%] Built target pcm_io                                                      
[ 88%] Built target flake_static                                                
[ 94%] Built target flake_exe                                                   
[100%] Built target wavinfo                                                     
Install the project...                                                          
-- Install configuration: "Release"                                             
-- Installing: /usr/local/bin/flake                                             
-- Installing: /usr/local/bin/wavinfo                                           
-- Installing: /usr/local/include/flake/flake.h

The problem is that there it can't read the wav files correctly. The same happens with AIFF Files.
Code: [Select]
root@freetibet:/home/frank/flake-enc/build# wavinfo '/home/frank/Musik/Dissecting Table - Groping in the Dark/02 - Dissecting Table - Root of Evil.wav'
invalid RIFF id in wav header

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
File:
   Name:          /home/frank/Musik/Dissecting Table - Groping in the Dark/02 -Dissecting Table - Root of Evil.wav
   File Size:     34219292
Format:
   Type:          Microsoft Unknown Wave Format
   Channels:      0
   Sample Rate:   0 Hz
   Avg bytes/sec: 0
   Block Align:   0 bytes
   Bit Width:     0
Data:
   Start:         0
   Data Size:     0
   Leftover:  34219292 bytes
   Samples:       unknown
   Playing Time:  unknown
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Apart from that I've made a working windows 32bit build using the MinGW Crosscompiler of FreeBSD.
Flake_mingw32_i386.zip
WavPack 4.50.1 -hhx6 | LAME 3.98.2 -V 0

Flake vs Flac and the -Ax2 switch

Reply #23 –
Quote from: vpa on
Ooops ... The PPC Mac OS X builds aren't working! 
I was noticing this when when trying to build on  PPC Linux, that there are a lot warnings which don't appear on the other OSes (even not on my PPC Mac):

Those warnings occur with recent GCC versions.  I've looked into them and there are no issues.

Quote
The problem is that there it can't read the wav files correctly. The same happens with AIFF Files.

Weird.  Is that with all WAV and AIFF files or only some?  I might be able to find some access a PPC Linux machine to investigate.

Flake vs Flac and the -Ax2 switch

Reply #24 –
Yes, it's happening with all AIFF and Wav Files I tried. Maybe it's some kind of Big Endian Bug in libpcm_io? Then also Sparc, MIPS & MC68000 Machines will be affected.
I might try to make an Amiga crosscompile under FreeBSD to see if it only affects PPC or all Big Endian architectures.
I've made a SVN build on my PPC iMac maybe a year ago and that one works without problems. That's why I didn't fully test it in first place.

Another weird thing:
With very noisy music (Death Metal, Japanese Noise, Industrial) the compression ratio is like this:
[best] flake -12 > flac -8 -A tukey\(0.5\) -A flattop > flac -8 > flake -8 [worst]
but with very silent music (Piano Sonatas, String Quartets) it is like this:
[best] flac -8 -A tukey\(0.5\) -A flattop > flac -8 > flake -12 > flake -8 [worst]
WavPack 4.50.1 -hhx6 | LAME 3.98.2 -V 0