TAK 2.2.0 - Alpha release Options

[TBeck]

TAK 2.2.0 Final

has been released.

[Steve Forte Rio]

surprisingly!
Very good news!
Thank you!

[CoRoNe]

Therefore i am very interested into compression results and comparisons with other codecs.

I have done some testing:

CODE

Chris Rea_test (Driving Home For Christmas sample)
WAV (16bit 44.1KHz, 11.733s)                                             1411Kbps, 2021.23KB, 100%
FLAC 1.2.1          [-8]                                                  861Kbps, 1233.17KB, 61.01%
TTA 3.4.1           [-e]                                                  852Kbps, 1220.10KB, 60.36%
WavPack 4.60.1      [-hhx]                                                837Kbps, 1199.49KB, 59.34%
TAK 2.0.0           [-e -p4m]                                             812Kbps, 1163.19KB, 57.55%
TAK 2.1.0           [-e -p4m]                                             812Kbps, 1163.02KB, 57.54%
TAK 2.2.0           [-e -p4m]                                             812Kbps, 1163.10KB, 57.54%
Monkey's Audio 4.10 [-c3000]                                              819Kbps, 1172.99KB, 58.03%
Monkey's Audio 4.10 [-c4000]                                              804Kbps, 1151.77KB, 56.98%
Monkey's Audio 4.10 [-c5000]                                              796Kbps, 1140.78KB, 56.44%
OptimFROG 4.600ex   [--encode --mode bestnew --seek min --optimize best]  791Kbps, 1133.00KB, 56.05%
OptimFROG 4.910b    [--encode --mode bestnew --seek min --optimize best]  790Kbps, 1131.81KB, 56.00%
OptimFROG 4.600ex   [--maximumcompression --experimental] (decode error!) 788Kbps, 1129.07KB, 55.86%
OptimFROG 4.910b    [--maximumcompression --experimental] (±55% cpu lol!) 787Kbps, 1127.71KB, 55.79%
LA 0.4b             [-high]                                               790Kbps, 1130.99KB, 55.96%

Gladiator Soundtrack
WAV (16bit 44.1KHz, 1h:01m:38.293s)     1411Kbps, 637088.86KB (622.16MB), 100%
FLAC 1.2.1          [--best]             706Kbps, 318757.12KB (311.29MB), 50.03%
WavPack 4.60.1      [-hhx]               702Kbps, 317369.66KB (309.93MB), 49.82%
TAK 2.1.0           [-e -p4m]            684Kbps, 308839.48KB (301.60MB), 48.48%
TAK 2.2.0           [-e -p4m]            684Kbps, 308865.17KB (301.63MB), 48.48%
Monkey's Audio 4.10 [-c3000]             687Kbps, 310348.86KB (303.08MB), 48.71%
Monkey's Audio 4.10 [-c4000]             680Kbps, 307066.54KB (299.87MB), 48.20%
Monkey's Audio 4.10 [-c5000]             676Kbps, 305098.90KB (297.95MB), 47.89%

Angra - Temple of Shadows
WAV (16bit 44.1KHz, 1h:06m:34.533s)     1411Kbps, 688120.82KB (671.99MB), 100%
FLAC 1.2.1          [--best]             986Kbps, 480797.87KB (469.53MB), 69.87%
WavPack 4.60.1      [-hhx]               971Kbps, 473486.58KB (462.39MB), 68.81%
TAK 2.1.0           [-e -p4m]            957Kbps, 466692.68KB (455.75MB), 67.82%
TAK 2.2.0           [-e -p4m]            957Kbps, 466720.24KB (455.78MB), 67.83%
Monkey's Audio 4.10 [-c3000]             960Kbps, 468348.68KB (457.37MB), 68.06%
Monkey's Audio 4.10 [-c4000]             952Kbps, 464030.97KB (453.16MB), 67.43%
Monkey's Audio 4.10 [-c5000]             948Kbps, 462497.08KB (451.66MB), 67.21%
LA 0.4b             [-high]              939Kbps, 457720.45KB (446.99MB), 66.52%

Royal Hunt - Paper Blood
WAV (16bit 44.1KHz, 56m:51.027s)        1411Kbps, 587602.68KB (573.83MB), 100%
FLAC 1.2.1          [--best]            1097Kbps, 456942.46KB (446.23MB), 77.76%
WavPack 4.60.1      [-hhx]              1089Kbps, 453645.32KB (443.01MB), 77.20%
TAK 2.1.0           [-e -p4m]           1082Kbps, 450697.51KB (440.13MB), 76.70%
TAK 2.2.0           [-e -p4m]           1083Kbps, 450738.92KB (440.17MB), 76.71%
Monkey's Audio 4.10 [-c3000]            1081Kbps, 449852.87KB (439.31MB), 76.56%
Monkey's Audio 4.10 [-c4000]            1078Kbps, 448739.60KB (438.22MB), 76.37%
Monkey's Audio 4.10 [-c5000]            1075Kbps, 447681.16KB (437.19MB), 76.19%

...what may on average result in about 0.01 percent worse compression.

I come to the same conclusion, although I would've loved seeing some more compression strength. The codec is already fast enough imho.
One funny detail about about my results; TAK -e -p4m beats Monkey's Audio 'High' in all cases except the last one (1082 vs 1081Kbps). Although my lossless collection is still rather small, Paper Blood appeared to be the hardest to compress.

The inclusion of the cuesheet should already work, although not with the '*' placeholder:

CODE

-tt #      Add textual tag item #, where # is a key/value pair: "key=value",
           for instance "TITLE=A nice song". "key=@file" will read the value
           from the text(!) file "file" in the source directory.

Takc.exe -e -p4m -tt "Cuesheet=@filename.cue" <infile> <outfile.tak> doesn't work. Neither when I include the full directory path.
"Command line error: Error reading tag file" is what I get.

[_m²_]

Updated my test with TTA, WavPack, TAK.
Generally, TAK looses to ALS, but is rocket fast.
I haven't decided which one do I want to use yet...Really, I expected TAK to do better.
http://www.hydrogenaudio.org/forums/index....showtopic=89649

[TBeck]

Updated my test with TTA, WavPack, TAK.

I took the freedom to post an excerpt of your data in the final release thread. I hope, that's ok?

[TBeck]

I have done some testing:

Thank you!

...what may on average result in about 0.01 percent worse compression.

I come to the same conclusion, although I would've loved seeing some more compression strength. The codec is already fast enough imho.

I don't think i can significantly improve the compression without making decoding slower.

One funny detail about about my results; TAK -e -p4m beats Monkey's Audio 'High' in all cases except the last one (1082 vs 1081Kbps). Although my lossless collection is still rather small, Paper Blood appeared to be the hardest to compress.

As a rule of thumb: The harder files are to compress, the worse TAK will perform compared to MAC. When it comes to highly compressible files, usually dynamically rich and tonal, TAK can do even better than MAC Extra, as can be seen in the FLAC comparison.

The inclusion of the cuesheet should already work, although not with the '*' placeholder:

Takc.exe -e -p4m -tt "Cuesheet=@filename.cue" <infile> <outfile.tak> doesn't work. Neither when I include the full directory path.
"Command line error: Error reading tag file" is what I get.

Strange. Where is the cue file located? In the source directory?

[_m²_]

Updated my test with TTA, WavPack, TAK.

I took the freedom to post an excerpt of your data in the final release thread. I hope, that's ok?

Sure

[CoRoNe]

Strange. Where is the cue file located? In the source directory?

Yes, in the source directory. But I think I figured it out. Initially I used the following command line:

CODE

wvunpack <infile.wv> - | Takc -e -p4m -tt "Cuesheet=@infile.cue" - <outfile.tak>

Which results in the error I mentioned before.
When I first decode the wv-file and use the wav-file as source it does seem to work. So it appears the TAK encoder can't load the cue-file properly when being used in a pipe.
When I then insert the tak-file in Foobar however, it doesn't load the cuesheet at all. After examining the tak-file in MP3tag there appear to be two cuesheet-entries. A cuesheet-entry with the correct content, but also a blank one (with no content). Manually deleting the blank cuesheet-entry in MP3tag does the trick, but still there appears to be something going wrong in the process.

I don't think i can significantly improve the compression without making decoding slower.

As long as it doesn't become as slow as Monkey's Audio 'Insane' I really don't mind some degeneration. I think a lot more people wouldn't mind either. Someone should hold a poll some time.

This post has been edited by CoRoNe: Jul 13 2011, 17:46

[TBeck]

Thank you for the info! I will look into it for the next release. Probably very few (if any) users have tried TAK's tagging functionality, what might explain why this bug has not been reported earlier.

This post has been edited by TBeck: Jul 13 2011, 17:41