Hydrogenaudio Forums > Improved FLAC encoder

Hydrogenaudio Forums > Lossless Audio Compression > FLAC

Pages: 1, 2

jcoalson

May 2 2006, 19:30

Click to view attachment
This is a win32 version of the flac-1.1.2 encoder with improvements based on Garf's suggestions (or it can be rebuilt from the current CVS code). For me on CD audio, properly applied it can improve the compression ratio 1% or 2% (e.g. ratio dropping by 1% from 51% to 50%, or a 2% improvement on a ratio of ~50%) with no effect on the decoding speed or complexity.

Garf's suggestion about applying a Hann window to the data before LPC analysis prompted me to go back to that part of the codec and investigate what could be improved. Some of the things normally done in LPC analysis (like overlapping windows) would be a big pain to do in the reference encoder because of the way it's designed (and probably with little gain). Other things (like adjusting for the mean in autocorrelation) are easy but yield little gain. But using one or more windows on the data has had an immediately positive effect.

This version of flac adds a new -A option that lets you specify one or more window functions. Instead of a rectangular window, the default is currently tukey(0.5) (first and last quarters of the window are cosine-tapered). Here is how it's used:

CODE

-A "function", --apodization="function"

Window audio data with given the apodization function. The functions are: bartlett, bartlett_hann, blackman, blackman_harris_4term_92db, connes, flattop, gauss(STDDEV), hamming, hann, kaiser_bessel, nuttall, rectangle, triangle, tukey(P), welch.

For gauss(STDDEV), STDDEV is the standard deviation (0<STDDEV<=0.5).

For tukey(P), P specifies the fraction of the window that is tapered (0<=P<=1; P=0 corresponds to "rectangle" and P=1 corresponds to "hann").

More than one -A option (up to 32) may be used. Any function that is specified erroneously is silently dropped. The encoder chooses suitable defaults in the absence of any -A options; any -A option specified replaces the default(s).

***When more than one function is specified, then for every subframe the encoder will try each of them separately and choose the window that results in the smallest compressed subframe. Multiple functions can greatly increase the encoding time.***

What I would like help with is coming up with optimal general-case -A defaults for the -3 .. -8 presets. In my testing the optimal window seems to depend heavily on the spectral characteristics of the audio (as you would expect). If you use multiple windows that have different characteristics this leads to better overall results but increases the encoding time. So what I'd like to answer is:

1. What is the best window for CD audio "in general" (to be used for the lower presets)
2. What are the best 2 (and maybe 3 if it makes a big difference) windows (to be used for the highest presets)

This can be done with with brute force (try all combinations), heuristics (by knowing the characteristics of each window) or haphazard. I'm hoping that with your help we'll be able to arrive at general answers. I'm afraid mine are too specific and biased by my corpus.

So if anyone who wants to experiment can try with different -A settings, and reply here with their results (compression differences and the nature of the corpus like "classical", "metal", "noisy", "quiet") that would be great. Thanks to Garf and thanks in advance to testers. I'll post my results with the standard FLAC corpus next.

Josh

these are the results I got trying each window in turn with -5 and -8.

-5 uncompressed size=780558856

CODE

409849350 tukey(0.125)
409855256 tukey(0.25)
409875833 tukey(0.03125)
409877499 tukey(0.0625)
409883425 tukey(0.015625)
409892304 tukey(0.5)
409896700 tukey(0.0078125)
409919399 gauss(0.5)
409942585 hamming
410099774 welch
410127063 triangle
410139341 connes
410145159 hann
410206946 blackman
410209682 kaiser_bessel
410245892 blackman_harris_4term_92db
410261735 nuttall
410334114 gauss(0.25)
410443236 flattop
410593346 gauss(0.125)
411043098 gauss(0.0625)
411764264 gauss(0.03125)
413032246 bartlett_hann
413099455 gauss(0.015625)
413457591 rectangle
413871409 bartlett
415302473 gauss(0.0078125)

-5 for the "non-classical" tracks (which is what I think the general populace is biased toward)
uncompressed size=321208244

CODE

213272941 hamming
213279899 gauss(0.5)
213293852 tukey(0.5)
213324056 connes
213326549 welch
213330829 triangle
213334318 tukey(0.25)
213337944 hann
213351475 blackman
213362011 kaiser_bessel
213362020 tukey(0.125)
213377272 tukey(0.0625)
213383367 nuttall
213383604 blackman_harris_4term_92db
213388948 tukey(0.03125)
213396140 tukey(0.015625)
213400261 tukey(0.0078125)
213431582 gauss(0.25)
213457420 flattop
213615243 gauss(0.125)
213864173 gauss(0.0625)
214223504 gauss(0.03125)
214376512 bartlett_hann
214445900 rectangle
214763174 bartlett
214836383 gauss(0.015625)
215909945 gauss(0.0078125)

-8 uncompressed size=780558856

CODE

406740688 tukey(0.5)
406786760 tukey(0.25)
406824594 tukey(0.125)
406850821 welch
406862118 tukey(0.0625)
406863285 hamming
406879377 connes
406879491 hann
406884072 tukey(0.03125)
406891990 tukey(0.015625)
406897041 tukey(0.0078125)
406897975 triangle
406929514 blackman
406943550 kaiser_bessel
406949271 blackman_harris_4term_92db
406971792 nuttall
407010949 gauss(0.25)
407144075 gauss(0.5)
407155852 flattop
407198841 gauss(0.125)
407519399 gauss(0.0625)
408079300 gauss(0.03125)
409140225 gauss(0.015625)
410968550 gauss(0.0078125)
411083716 bartlett_hann
411877779 rectangle
412069308 bartlett

-8 "non-classical" uncompressed size=321208244

CODE

212016651 tukey(0.5)
212020858 hamming
212025289 hann
212026935 connes
212035402 blackman
212035638 triangle
212037703 welch
212045149 kaiser_bessel
212053628 blackman_harris_4term_92db
212055529 nuttall
212061598 tukey(0.25)
212081934 gauss(0.25)
212087652 gauss(0.5)
212091180 tukey(0.125)
212114878 tukey(0.0625)
212126112 tukey(0.03125)
212130608 flattop
212132781 tukey(0.015625)
212136244 tukey(0.0078125)
212212269 gauss(0.125)
212411636 gauss(0.0625)
212734806 gauss(0.03125)
213284875 gauss(0.015625)
213419896 bartlett_hann
213599770 rectangle
213869540 bartlett
214205800 gauss(0.0078125)

goodnews

May 2 2006, 19:41

While I appreciate you trying to "tweak" a 1-2% compression improvement out of FLAC, I sure hope it doesn't break backward compatability with FLAC decoders or audio players that are out there. I like FLAC just the way it is (a lossless standard) that just works. I have entrusted my entire archive of audio masters to the FLAC format and would dislike immensely having to deal with a new version that might be buggy and/or break compatability over a 1 to 2% coding improvement. My 2 cents... Thanks for writing FLAC.

Shade[ST]

May 2 2006, 21:01

In your case, M. Coalson (do you mind if I call you Sir?

), your best bet would be to pick half a turkey -- the default is fine as it is right now, and seems quite optimized. Maybe allowing a custom setting on the generic encoder could help people trying to squeeze a few drops of extra optimization out, but at least from the standard flac corpus, it seems to be optimized to the maximum, right now.

Short answer : Tukey 0.5

Is there any slowdown resulting of this window application (when it is changed)?

May 2 2006, 21:02

If I've observed one thing about the way Josh works over the years I've been following (and using!) FLAC, it's that he is careful to always preserve backward compatibility. No worries there....

- M.

Zoom

May 2 2006, 21:17

Yeah I did a quick test:

Quick Test

I only tested -8 because that's the only preset I use. I didn't really pour over the results, I just highlighted the setting that gave the best compression. I'll do some more files tomorrow, but as of right now it's already past my normal bedtime.

jcoalson

May 2 2006, 21:19

QUOTE

' date='May 2 2006, 10:01 PM' post='388693']In your case, M. Coalson (do you mind if I call you Sir?

)

hehe, am I being knighted?

I'm just a Josh...

QUOTE

Is there any slowdown resulting of this window application (when it is changed)?

no, only if you specify multiple -A options. even in the default case one window is always applied, and applying the window itself is pretty negligible compared to the rest of the analysis stage.

Josh

jcoalson

May 2 2006, 22:04

QUOTE(goodnews @ May 2 2006, 08:41 PM)

While I appreciate you trying to "tweak" a 1-2% compression improvement out of FLAC, I sure hope it doesn't break backward compatability with FLAC decoders or audio players that are out there.

don't worry, this has no effect on the decoder, it's purely an encoder improvement.

@zoom, thanks, that format is useful.

also, a couple other things that would be interesting to know... 1. if there there are any mistakes in any of the window formulas (wouldn't surprise me)

Click to view attachment

2. if there is an efficient technique for determining what kind of window would be best. I imagine that looking at the signal in the frequency domain might give a hint but I'm not sure that would be faster than just trying a couple of different windows.

Josh

Egor

May 3 2006, 01:04

Windows users can use this .cmd file to test the encoder:

CODE

for %%n in ("tukey(0.875)" "tukey(0.75)" "tukey(0.625)" "tukey(0.5)" "tukey(0.375)" "tukey(0.25)" "tukey(0.125)" "gauss(0.5)" "gauss(0.4375)" "gauss(0.375)" "gauss(0.25)" "gauss(0.1875)" "gauss(0.125)" "gauss(0.0625)" "hamming" "hann" "blackman" "kaiser_bessel" "connes" "triangle" "welch" "blackman_harris_4term_92db" "nuttall" "flattop" "bartlett_hann" "rectangle" "bartlett") do (start "%%n" /B /LOW /WAIT "flac.exe" -8 -A %%n "test.wav" -o "%%n.flac" & dir /N "*.flac" >>"temp.txt" & del "*.flac")
dir /N *.wav | find ".wav" >"report.txt"
type temp.txt | find ".flac" | sort >>"report.txt"

or this one for users with "," (comma) as decimal point in system locale (e.g., Russia):

CODE

for %%n in ("tukey(0,875)" "tukey(0,75)" "tukey(0,625)" "tukey(0,5)" "tukey(0,375)" "tukey(0,25)" "tukey(0,125)" "gauss(0,5)" "gauss(0,4375)" "gauss(0,375)" "gauss(0,25)" "gauss(0,1875)" "gauss(0,125)" "gauss(0,0625)" "hamming" "hann" "blackman" "kaiser_bessel" "connes" "triangle" "welch" "blackman_harris_4term_92db" "nuttall" "flattop" "bartlett_hann" "rectangle" "bartlett") do (start "%%n" /B /LOW /WAIT "flac.exe" -8 -A %%n "test.wav" -o "%%n.flac" & dir /N "*.flac" >>"temp.txt" & del "*.flac")
dir /N *.wav | find ".wav" >"report.txt"
type temp.txt | find ".flac" | sort >>"report.txt"

Copy contents to a .cmd file, place it in a folder with the new flac.exe and some test.wav and run. Output files will be named as blackman.flac etc., report.txt file will contain sorted (ascending sizes) list of files.

Edit. Updated the script to run with IDLE priority (in backgroung) and modified list of functions.
Edit 2. Made the script not to create bunch of flac files.

SebastianG

May 3 2006, 01:28

Are the blocks' windows currently overlapping?
Did you try it Vorbis-style? (overlapping + Vorbis window)

Sebi

Insolent

May 3 2006, 03:02

I encoded a few songs using Egor's .cmd. tukey(0.25) seems to give better results for the heavier songs

Pink Floyd - Time
Prog Rock

CODE

74,946,524     original wav

43,298,415     tukey(0.5)
43,299,286     connes
43,299,513     triangle
43,301,660     hann
43,302,460     welch
43,303,237     blackman
43,303,325     tukey(0.25)
43,303,921     kaiser_bessel
43,305,526     blackman_harris_4term_92db
43,308,517     nuttall
43,310,975     tukey(0.125)
43,311,345     gauss(0.25)
43,312,623     hamming
43,314,715     tukey(0.0625)
43,319,185     flattop
43,319,745     tukey(0.03125)
43,320,245     tukey(0.015625)
43,320,473     tukey(0.0078125)
43,333,742     gauss(0.125)
43,347,970     gauss(0.5)
43,375,339     gauss(0.0625)
43,449,666     gauss(0.03125)
43,574,439     gauss(0.015625)
43,772,976     gauss(0.0078125)
43,824,232     bartlett_hann
43,899,492     rectangle
43,952,327     bartlett

Ennio Morricone - The Ecstacy of Gold
Err... classical? :/

CODE

35,849,228     original wav

19,399,369     tukey(0.0078125)
19,399,423     tukey(0.03125)
19,399,551     tukey(0.015625)
19,399,588     tukey(0.0625)
19,400,363     tukey(0.125)
19,400,453     tukey(0.25)
19,401,527     tukey(0.5)
19,404,046     welch
19,406,528     triangle
19,406,895     hann
19,407,121     connes
19,409,607     hamming
19,410,162     kaiser_bessel
19,410,872     blackman
19,411,842     blackman_harris_4term_92db
19,412,074     nuttall
19,414,466     gauss(0.25)
19,415,691     gauss(0.5)
19,417,452     flattop
19,425,566     gauss(0.125)
19,438,696     gauss(0.0625)
19,462,919     gauss(0.03125)
19,507,293     gauss(0.015625)
19,590,546     gauss(0.0078125)
19,687,589     bartlett_hann
19,731,890     rectangle
19,745,621     bartlett

Symphony X - Inferno (Unleash The Fire)
Prog/Speed Metal

CODE

58,689,500     original wav

41,504,232     tukey(0.25)
41,505,556     tukey(0.5)
41,505,871     welch
41,506,215     tukey(0.125)
41,506,506     tukey(0.0625)
41,506,549     tukey(0.0078125)
41,507,788     tukey(0.03125)
41,507,834     tukey(0.015625)
41,513,590     triangle
41,516,383     connes
41,516,849     hamming
41,520,538     hann
41,522,831     gauss(0.5)
41,531,739     blackman
41,535,021     kaiser_bessel
41,542,636     blackman_harris_4term_92db
41,543,320     nuttall
41,554,387     gauss(0.25)
41,573,664     flattop
41,600,210     gauss(0.125)
41,641,811     gauss(0.0625)
41,694,697     gauss(0.03125)
41,716,091     bartlett_hann
41,730,106     rectangle
41,784,002     gauss(0.015625)
41,793,242     bartlett
41,920,042     gauss(0.0078125)

In Flames - Acoustic Medley
Instrumental (acoustic harmony)

CODE

27,076,268     original wav

13,895,201     gauss(0.125)
13,896,080     gauss(0.0625)
13,904,998     gauss(0.25)
13,905,189     nuttall
13,905,232     flattop
13,905,571     blackman_harris_4term_92db
13,909,832     kaiser_bessel
13,910,960     blackman
13,913,454     hann
13,913,926     gauss(0.03125)
13,916,395     connes
13,917,893     triangle
13,922,512     welch
13,923,928     tukey(0.5)
13,932,504     hamming
13,934,215     tukey(0.25)
13,937,193     tukey(0.125)
13,940,294     tukey(0.0625)
13,941,648     tukey(0.03125)
13,944,015     tukey(0.015625)
13,948,647     tukey(0.0078125)
13,952,120     gauss(0.015625)
13,975,455     gauss(0.5)
14,013,929     gauss(0.0078125)
14,144,818     bartlett_hann
14,163,763     bartlett
14,192,943     rectangle

Enforsaken - Tales of Bitterness
Melodic Death Metal (heaviest song encoded)

CODE

61,542,476     original wav

46,212,870     tukey(0.25)
46,213,819     tukey(0.125)
46,215,229     tukey(0.0625)
46,215,631     tukey(0.5)
46,215,752     tukey(0.03125)
46,216,228     tukey(0.015625)
46,216,760     tukey(0.0078125)
46,217,205     welch
46,224,433     gauss(0.5)
46,225,065     triangle
46,225,543     hamming
46,227,927     connes
46,231,745     hann
46,242,472     blackman
46,245,556     kaiser_bessel
46,252,841     nuttall
46,253,016     blackman_harris_4term_92db
46,262,017     gauss(0.25)
46,278,066     flattop
46,308,016     gauss(0.125)
46,363,746     gauss(0.0625)
46,443,073     bartlett_hann
46,446,791     gauss(0.03125)
46,451,070     rectangle
46,537,101     bartlett
46,583,945     gauss(0.015625)
46,800,927     gauss(0.0078125)

Egor

May 3 2006, 04:37

It seems to me that testing with CD images is not practical: it takes very long to complete and it may contain long inter-track digital silence parts. So probably it is better to test with ONE song per test.

CODE

Michael Buble - It's time (Jazz Vocal / Popular)
flac -8

590 281 484 test.wav

367 258 943 tukey(0,625)
367 260 079 tukey(0,618034) 
367 264 606 tukey(0,5)
367 269 016 tukey(0,75)
367 284 590 tukey(0,875)
367 292 956 connes
367 293 894 hamming
367 293 912 welch
367 301 748 hann
367 302 185 triangle
367 307 853 tukey(0,25)
367 333 950 blackman
367 346 199 kaiser_bessel
367 348 342 tukey(0,125)
367 360 482 gauss(0,5)
367 366 460 nuttall
367 369 648 blackman_harris_4term_92db
367 381 199 tukey(0,0625)
367 400 137 tukey(0,03125)
367 407 823 gauss(0,25)
367 410 875 tukey(0,015625)
367 414 333 tukey(0,0078125)
367 485 735 flattop
367 628 021 gauss(0,125)
367 934 476 gauss(0,0625)
368 439 444 gauss(0,03125)
369 179 530 bartlett_hann
369 305 745 gauss(0,015625)
369 486 263 rectangle
369 771 271 bartlett
370 751 786 gauss(0,0078125)

Mo0zOoH

May 3 2006, 05:24

Seems like tukey (0.5) is an overall champion.

Egor

May 3 2006, 09:11

Genre classification is taken from Amazon.com ("Look for similar items by category"). All links lead to corresponding Audio CD item with short 20-30 sec. music clips available for listen in WMA and/or RA.

CODE

Seal - Lost My Faith
Adult Contemporary / Pop / Pop Rock / Vocal Pop / Rock

48 474 764 test.wav

29 797 004 tukey(0,75)
29 797 180 tukey(0,6875)
29 797 213 tukey(0,25)
29 797 458 tukey(0,125)
29 798 177 tukey(0,71875)
29 798 222 tukey(0,625)
29 798 398 tukey(0,5)
29 800 109 tukey(0,0625)
29 801 244 tukey(0,03125)
29 802 229 tukey(0,015625)
29 803 137 tukey(0,875)
29 804 148 welch
29 804 438 hamming
29 804 873 connes
29 805 012 tukey(0,0078125)
29 806 527 triangle
29 807 314 hann
29 809 327 blackman
29 811 448 kaiser_bessel
29 811 496 nuttall
29 812 210 blackman_harris_4term_92db
29 813 799 gauss(0,5)
29 817 479 gauss(0,25)
29 824 973 flattop
29 830 982 gauss(0,125)
29 860 623 gauss(0,0625)
29 909 657 gauss(0,03125)
29 990 596 gauss(0,015625)
30 033 688 bartlett_hann
30 067 595 rectangle
30 099 657 bartlett
30 132 632 gauss(0,0078125)

CODE

Bryan Adams - Cloud Number Nine
Pop / Pop Rock / Rock / Adult Contemporary

39 788 828 test.wav

27 354 308 blackman
27 354 541 gauss(0,25)
27 354 698 blackman_harris_4term_92db
27 355 443 nuttall
27 356 279 flattop
27 356 694 kaiser_bessel
27 357 733 tukey(0,875)
27 358 142 hann
27 358 818 connes
27 359 964 hamming
27 360 192 tukey(0,71875)
27 360 643 tukey(0,75)
27 360 721 tukey(0,6875)
27 361 506 gauss(0,125)
27 363 033 triangle
27 363 394 tukey(0,625)
27 366 078 tukey(0,5)
27 366 474 welch
27 372 787 gauss(0,5)
27 375 991 gauss(0,0625)
27 377 582 tukey(0,25)
27 387 775 tukey(0,125)
27 393 651 tukey(0,0625)
27 396 525 tukey(0,03125)
27 397 724 tukey(0,015625)
27 398 791 tukey(0,0078125)
27 406 955 gauss(0,03125)
27 466 587 gauss(0,015625)
27 562 553 gauss(0,0078125)
27 564 003 bartlett_hann
27 601 758 rectangle
27 620 854 bartlett

CODE

Acoustic Alchemy - So Kylie
Jazz / Smooth Jazz / New Age

46 247 420 test.wav

31 082 004 flattop
31 084 092 gauss(0,25)
31 085 215 blackman_harris_4term_92db
31 085 631 nuttall
31 087 579 kaiser_bessel
31 088 416 blackman
31 089 980 gauss(0,125)
31 095 130 hann
31 097 849 connes
31 097 851 tukey(0,875)
31 099 848 hamming
31 102 273 triangle
31 102 425 tukey(0,75)
31 104 121 tukey(0,71875)
31 105 341 tukey(0,6875)
31 108 436 tukey(0,625)
31 108 965 gauss(0,0625)
31 111 493 gauss(0,5)
31 114 695 welch
31 115 786 tukey(0,5)
31 136 227 tukey(0,25)
31 143 722 gauss(0,03125)
31 148 950 tukey(0,125)
31 156 207 tukey(0,0625)
31 160 024 tukey(0,03125)
31 161 325 tukey(0,015625)
31 162 139 tukey(0,0078125)
31 172 280 bartlett_hann
31 189 186 bartlett
31 203 537 gauss(0,015625)
31 209 187 rectangle
31 297 414 gauss(0,0078125)

CODE

3rd Force - You Got It
Jazz / New Age

48 423 020 test.wav

32 524 489 blackman_harris_4term_92db
32 524 503 nuttall
32 524 574 kaiser_bessel
32 524 638 blackman
32 525 499 gauss(0,25)
32 528 466 hann
32 529 867 flattop
32 530 914 tukey(0,875)
32 531 204 connes
32 534 237 tukey(0,75)
32 535 370 hamming
32 535 886 tukey(0,71875)
32 537 112 tukey(0,6875)
32 537 227 triangle
32 539 904 tukey(0,625)
32 542 725 gauss(0,125)
32 545 130 gauss(0,5)
32 546 652 welch
32 547 263 tukey(0,5)
32 565 544 gauss(0,0625)
32 567 758 tukey(0,25)
32 581 849 tukey(0,125)
32 590 546 tukey(0,0625)
32 594 973 tukey(0,03125)
32 597 231 tukey(0,015625)
32 597 751 tukey(0,0078125)
32 598 698 gauss(0,03125)
32 603 189 bartlett_hann
32 606 696 bartlett
32 648 365 rectangle
32 653 773 gauss(0,015625)
32 741 336 gauss(0,0078125)

CODE

The Temptations - A love I can see
Oldies / Classic R&B / R&B / Soul

61 039 148 test.wav

42 603 149 tukey(0,6875)
42 603 260 tukey(0,625)
42 604 157 tukey(0,75)
42 605 036 tukey(0,5)
42 605 283 welch
42 605 489 tukey(0,875)
42 605 912 triangle
42 605 960 hamming
42 606 559 connes
42 608 149 hann
42 609 320 gauss(0,5)
42 613 505 tukey(0,25)
42 613 851 blackman
42 616 907 kaiser_bessel
42 622 192 nuttall
42 622 417 tukey(0,125)
42 622 759 blackman_harris_4term_92db
42 626 694 tukey(0,0625)
42 629 175 tukey(0,03125)
42 630 495 tukey(0,015625)
42 631 024 tukey(0,0078125)
42 631 272 gauss(0,25)
42 645 378 flattop
42 672 991 gauss(0,125)
42 721 847 gauss(0,0625)
42 767 332 bartlett_hann
42 772 875 rectangle
42 788 818 gauss(0,03125)
42 857 372 bartlett
42 895 465 gauss(0,015625)
43 059 477 gauss(0,0078125)

Synthetic Soul

May 3 2006, 09:26

Glad to see someone testing using tukey values over 0.5.

Josh may have only listed results from 0-0.5; however, as his docs state, it ranges from 0-1.

Egor's results appear to show that a higher value than 0.5 may be better...

Edit: I was actually hoping to be smug and be the first to test higher values, but Egor beat me to the punch.

Edit 2: Am I going mental? I thought the original batch script didn't include values over 0.5, and Insolent's results appear to bear that out. I dunno, I'm confused. Remember to test values over 0.5 anyway children, but always get an adult to supervise.

Cartman_Sr

May 3 2006, 09:50

I did a test just now with Egor's script and here's the results:
Would it be useful to record the length of time it takes to do each test for encoding and decoding? The only problem is it wouldn't be very accurate because the guy working the stopwatch is a bit tired right now

CODE

File: Econoline Crush "Surefire". Moderately fast, hard rock,
album replay gain value is -7.18, peak level 0.982.
Track 01 in album "The Devil You Know".

Original size of .wav: 37996604

Baseline test with the old flac.exe 1.1.2 with -8 results: size: 26548249, ratio: 0.699
Baseline test with the old flac.exe 1.1.2 with -5 results: size: 26597329, ratio: 0.700

    26,373,390    tukey(0.0625).flac
    26,373,419    tukey(0.015625).flac
    26,373,552    tukey(0.03125).flac
    26,373,866    tukey(0.25).flac
    26,374,304    tukey(0.0078125).flac
    26,375,957    tukey(0.5).flac
    26,376,569    tukey(0.625).flac
    26,376,703    welch.flac
    26,379,314    tukey(0.75).flac
    26,381,420    tukey(0.875).flac
    26,381,883    connes.flac
    26,382,157    triangle.flac
    26,383,403    hamming.flac
    26,383,762    hann.flac
    26,384,467    gauss(0.5).flac
    26,389,232    blackman.flac
    26,390,587    kaiser_bessel.flac
    26,393,578    nuttall.flac
    26,394,967    blackman_harris_4term_92db.flac
    26,398,926    gauss(0.25).flac
    26,406,505    flattop.flac
    26,419,611    gauss(0.125).flac
    26,448,924    gauss(0.0625).flac
    26,494,724    gauss(0.03125).flac
    26,537,577    bartlett_hann.flac
    26,548,247    rectangle.flac
    26,572,386    gauss(0.015625).flac
    26,595,903    bartlett.flac
    26,689,265    gauss(0.0078125).flac

Cartman_Sr

May 3 2006, 10:47

Here's another one:

CODE

File: Black Sabbath "Reunion", track 7 of disc 2, "Paranoid (live)". Metal, with a
loud audience track.
.wav file size: 47604524 bytes
Baseline tests with old flac 1.1.2: -8 result was 33483792 bytes, ratio 0.703
                   -5 result was 33539684 bytes, ratio 0.705

        33,350,660    tukey(0.25).flac
        33,350,752    tukey(0.125).flac
        33,351,083    tukey(0.03125).flac
        33,351,089    tukey(0.0625).flac
        33,351,553    tukey(0.015625).flac
        33,351,644    tukey(0.0078125).flac
        33,353,213    tukey(0.5).flac
        33,354,387    welch.flac
        33,355,954    tukey(0.625).flac
        33,358,746    tukey(0.75).flac
        33,360,615    gauss(0.5).flac
        33,361,061    triangle.flac
        33,361,724    tukey(0.875).flac
        33,362,022    connes.flac
        33,362,137    hamming.flac
        33,364,682    hann.flac
        33,371,447    blackman.flac
        33,373,167    kaiser_bessel.flac
        33,377,505    nuttall.flac
        33,378,299    blackman_harris_4term_92db.flac
        33,384,174    gauss(0.25).flac
        33,396,431    flattop.flac
        33,417,378    gauss(0.125).flac
        33,458,057    gauss(0.0625).flac
        33,483,790    rectangle.flac
        33,486,829    bartlett_hann.flac
        33,516,644    gauss(0.03125).flac
        33,544,766    bartlett.flac
        33,607,746    gauss(0.015625).flac
        33,740,096    gauss(0.0078125).flac

jcoalson

May 3 2006, 11:43

QUOTE(SebastianG @ May 3 2006, 02:28 AM)

Are the blocks' windows currently overlapping?

no, this is really hard to do with the reference encoder due to its design.

QUOTE(SebastianG @ May 3 2006, 02:28 AM)

Did you try it Vorbis-style? (overlapping + Vorbis window)

you mean this one? no, haven't tried it, what is the motivation behind that function?

Josh

edit: just found this

Demetris

May 3 2006, 12:11

Results from 4 CD images. Compression level is -5. If CD images are of any use, I can repeat the runs for -8, and maybe add some more. I used Egor's batch file (thanks).

Rock, Indie rock

CODE

Franz Ferdinand - Franz Ferdinand ( Rock, Indie rock )

Settings     flac -5

411.000.284  wav

263.146.891  hamming
263.213.796  tukey(0,75)
263.219.469  tukey(0,875)
263.225.582  connes
263.229.532  gauss(0,5)
263.234.058  tukey(0,625)
263.234.129  hann
263.242.693  triangle
263.244.283  tukey(0,5)
263.244.351  welch
263.256.663  blackman
263.265.702  kaiser_bessel
263.294.433  nuttall
263.298.751  blackman_harris_4term_92db
263.312.349  tukey(0,25)
263.348.230  gauss(0,25)
263.368.586  tukey(0,125)
263.384.188  flattop
263.403.331  tukey(0,0625)
263.416.217  tukey(0,03125)
263.426.772  tukey(0,015625)
263.431.307  tukey(0,0078125)
263.624.620  gauss(0,125)
264.007.491  gauss(0,0625)
264.667.490  gauss(0,03125)
265.609.974  bartlett_hann
265.670.090  gauss(0,015625)
265.804.430  rectangle
266.350.898  bartlett
267.144.977  gauss(0,0078125)

Classical, Chamber

CODE

Bartok - The 6 String Quartets - Hagen Quartett - CD2 ( Classical, Chamber )

Settings     flac -5

792.388.844  wav

328.807.460  tukey(0,015625)
328.808.011  tukey(0,03125)
328.816.780  tukey(0,0625)
328.817.533  tukey(0,0078125)
328.824.978  tukey(0,125)
328.848.866  tukey(0,25)
328.907.717  tukey(0,5)
328.954.232  tukey(0,625)
328.990.946  welch
329.034.004  tukey(0,75)
329.119.309  tukey(0,875)
329.142.472  gauss(0,5)
329.159.989  hamming
329.165.919  connes
329.245.401  hann
329.275.609  flattop
329.291.402  triangle
329.462.428  blackman
329.520.377  kaiser_bessel
329.621.737  nuttall
329.635.319  blackman_harris_4term_92db
329.791.393  gauss(0,25)
330.368.669  gauss(0,125)
331.143.291  gauss(0,0625)
331.630.913  rectangle
332.065.047  bartlett
332.083.237  bartlett_hann
332.234.396  gauss(0,03125)
333.747.781  gauss(0,015625)
335.701.502  gauss(0,0078125)

Rock, Pop, Dance, Electronica

CODE

Madonna - Music ( Rock, Pop, Dance, Electronica )

Settings     flac -5

521.913.548  wav

339.256.667  flattop
339.292.763  kaiser_bessel
339.292.915  nuttall
339.297.501  blackman_harris_4term_92db
339.300.268  blackman
339.313.694  gauss(0,25)
339.326.053  hann
339.341.414  hamming
339.345.183  tukey(0,875)
339.354.898  connes
339.370.916  tukey(0,75)
339.407.096  triangle
339.414.757  tukey(0,625)
339.466.101  gauss(0,5)
339.466.944  tukey(0,5)
339.474.566  welch
339.597.401  gauss(0,125)
339.642.737  tukey(0,25)
339.759.687  tukey(0,125)
339.820.297  tukey(0,0625)
339.860.098  tukey(0,03125)
339.882.491  tukey(0,015625)
339.901.977  tukey(0,0078125)
340.067.700  gauss(0,0625)
340.791.923  gauss(0,03125)
341.228.752  bartlett_hann
341.406.438  rectangle
341.715.784  bartlett
341.955.434  gauss(0,015625)
343.911.423  gauss(0,0078125)

Jazz

CODE

The Dave Brubeck Quartet - Time out ( Jazz )

Settings     flac -5

410.447.564  wav

233.113.139  tukey(0,75)
233.113.646  tukey(0,625)
233.118.157  tukey(0,875)
233.118.799  connes
233.119.233  tukey(0,5)
233.120.528  hamming
233.123.485  hann
233.124.423  triangle
233.126.009  welch
233.129.824  gauss(0,5)
233.145.960  blackman
233.154.044  kaiser_bessel
233.156.273  tukey(0,25)
233.173.263  nuttall
233.175.633  blackman_harris_4term_92db
233.193.930  tukey(0,125)
233.205.649  gauss(0,25)
233.218.003  tukey(0,0625)
233.232.815  tukey(0,03125)
233.240.404  tukey(0,015625)
233.240.596  flattop
233.246.803  tukey(0,0078125)
233.374.878  gauss(0,125)
233.589.049  gauss(0,0625)
233.647.616  bartlett_hann
233.691.834  rectangle
233.939.049  gauss(0,03125)
233.947.922  bartlett
234.596.944  gauss(0,015625)
235.735.481  gauss(0,0078125)

Night Rain

May 3 2006, 12:23

Original WAV 46,223,900

05/03/2006 02:09 PM 29,229,288 blackman.flac
05/03/2006 02:09 PM 29,229,510 kaiser_bessel.flac
05/03/2006 02:09 PM 29,230,648 nuttall.flac
05/03/2006 02:09 PM 29,230,869 blackman_harris_4term_92db.flac
05/03/2006 02:09 PM 29,232,884 hann.flac
05/03/2006 02:09 PM 29,233,992 gauss(0.25).flac
05/03/2006 02:09 PM 29,235,303 tukey(0.875).flac
05/03/2006 02:09 PM 29,235,885 connes.flac
05/03/2006 02:09 PM 29,237,784 hamming.flac
05/03/2006 02:09 PM 29,240,103 flattop.flac
05/03/2006 02:09 PM 29,240,127 tukey(0.75).flac
05/03/2006 02:09 PM 29,240,700 triangle.flac
05/03/2006 02:09 PM 29,247,420 tukey(0.625).flac
05/03/2006 02:09 PM 29,253,185 gauss(0.5).flac
05/03/2006 02:09 PM 29,253,760 welch.flac
05/03/2006 02:09 PM 29,257,277 tukey(0.5).flac
05/03/2006 02:09 PM 29,265,523 gauss(0.125).flac
05/03/2006 02:09 PM 29,285,601 tukey(0.25).flac
05/03/2006 02:09 PM 29,301,418 gauss(0.0625).flac
05/03/2006 02:09 PM 29,305,862 tukey(0.125).flac
05/03/2006 02:09 PM 29,317,644 tukey(0.0625).flac
05/03/2006 02:09 PM 29,323,825 tukey(0.03125).flac
05/03/2006 02:09 PM 29,327,476 tukey(0.015625).flac
05/03/2006 02:09 PM 29,328,710 tukey(0.0078125).flac
05/03/2006 02:09 PM 29,348,962 gauss(0.03125).flac
05/03/2006 02:09 PM 29,397,980 bartlett_hann.flac
05/03/2006 02:09 PM 29,422,630 gauss(0.015625).flac
05/03/2006 02:09 PM 29,433,043 rectangle.flac
05/03/2006 02:09 PM 29,445,063 bartlett.flac
05/03/2006 02:14 PM 29,461,245 test.flac
05/03/2006 02:09 PM 29,539,707 gauss(0.0078125).flac

Cartman_Sr

May 3 2006, 16:01

Need any more tests?

Insolent

May 3 2006, 16:54

QUOTE(Synthetic Soul @ May 4 2006, 01:26 AM)

Edit 2: Am I going mental? I thought the original batch script didn't include values over 0.5, and Insolent's results appear to bear that out. I dunno, I'm confused. Remember to test values over 0.5 anyway children, but always get an adult to supervise.

I think Egor edited his script after I'd already done my tests. Because I don't remeber seeing 0.75 and 0.875 in there.

HisInfernalMajesty

May 3 2006, 19:18

I ran two tests and I have more on the way later

CODE

Opeth - The Drapery Falls
Death Metal

Original size: 115,342,124

05/03/2006  08:07p          84,925,538 tukey(0.625).flac
05/03/2006  08:07p          84,926,049 tukey(0.75).flac
05/03/2006  08:07p          84,927,584 tukey(0.5).flac
05/03/2006  08:07p          84,928,237 welch.flac
05/03/2006  08:07p          84,929,268 tukey(0.875).flac
05/03/2006  08:07p          84,929,275 triangle.flac
05/03/2006  08:07p          84,930,206 connes.flac
05/03/2006  08:07p          84,931,772 hann.flac
05/03/2006  08:07p          84,933,449 hamming.flac
05/03/2006  08:07p          84,938,885 tukey(0.25).flac
05/03/2006  08:07p          84,941,754 blackman.flac
05/03/2006  08:07p          84,946,453 kaiser_bessel.flac
05/03/2006  08:07p          84,951,194 tukey(0.125).flac
05/03/2006  08:07p          84,952,685 blackman_harris_4term_92db.flac
05/03/2006  08:07p          84,953,268 nuttall.flac
05/03/2006  08:07p          84,955,396 gauss(0.5).flac
05/03/2006  08:07p          84,956,352 tukey(0.0625).flac
05/03/2006  08:07p          84,961,363 tukey(0.03125).flac
05/03/2006  08:07p          84,964,095 tukey(0.015625).flac
05/03/2006  08:07p          84,964,670 tukey(0.0078125).flac
05/03/2006  08:07p          84,965,453 gauss(0.25).flac
05/03/2006  08:07p          84,987,934 flattop.flac
05/03/2006  08:07p          85,029,063 gauss(0.125).flac
05/03/2006  08:07p          85,106,610 gauss(0.0625).flac
05/03/2006  08:07p          85,225,779 gauss(0.03125).flac
05/03/2006  08:07p          85,413,339 bartlett_hann.flac
05/03/2006  08:07p          85,418,708 gauss(0.015625).flac
05/03/2006  08:07p          85,456,032 rectangle.flac
05/03/2006  08:07p          85,586,037 bartlett.flac
05/03/2006  08:07p          85,709,985 gauss(0.0078125).flac

And just out of interest, I ran a mono file through:

CODE

Phil Harris - "The Dark Town Poker Club" (mono)
Big Band

Original Size: 17,057,924

05/03/2006  08:27p           8,397,704 gauss(0.03125).flac
05/03/2006  08:27p           8,400,230 gauss(0.015625).flac
05/03/2006  08:27p           8,405,118 gauss(0.0625).flac
05/03/2006  08:27p           8,405,835 gauss(0.125).flac
05/03/2006  08:27p           8,406,988 tukey(0.125).flac
05/03/2006  08:27p           8,409,333 tukey(0.875).flac
05/03/2006  08:27p           8,410,243 tukey(0.5).flac
05/03/2006  08:27p           8,411,396 welch.flac
05/03/2006  08:27p           8,411,481 tukey(0.015625).flac
05/03/2006  08:27p           8,411,784 triangle.flac
05/03/2006  08:27p           8,412,951 blackman.flac
05/03/2006  08:27p           8,412,953 nuttall.flac
05/03/2006  08:27p           8,413,914 tukey(0.03125).flac
05/03/2006  08:27p           8,414,332 tukey(0.25).flac
05/03/2006  08:27p           8,414,512 gauss(0.25).flac
05/03/2006  08:27p           8,414,700 hann.flac
05/03/2006  08:27p           8,415,178 tukey(0.0078125).flac
05/03/2006  08:27p           8,415,700 tukey(0.0625).flac
05/03/2006  08:27p           8,416,148 tukey(0.625).flac
05/03/2006  08:27p           8,416,680 kaiser_bessel.flac
05/03/2006  08:27p           8,417,520 tukey(0.75).flac
05/03/2006  08:27p           8,417,637 blackman_harris_4term_92db.flac
05/03/2006  08:27p           8,420,836 connes.flac
05/03/2006  08:27p           8,423,927 flattop.flac
05/03/2006  08:27p           8,436,133 gauss(0.0078125).flac
05/03/2006  08:27p           8,445,898 hamming.flac
05/03/2006  08:27p           8,503,197 gauss(0.5).flac
05/03/2006  08:27p           8,871,029 bartlett_hann.flac
05/03/2006  08:27p           8,937,868 bartlett.flac
05/03/2006  08:27p           8,958,712 rectangle.flac

Egor

May 3 2006, 21:56

QUOTE(Synthetic Soul @ May 3 2006, 10:26 PM)

I added values over 0.5 a little later, sorry for not mentioning it in the original post.

A_Man_Eating_Duck

May 4 2006, 01:27

A Perfect Circle - Mer De Noms (full Album)
Style: Hard Rock

CODE

Original Wav
471,199,724 A Perfect Circle - Mer De Noms.wav

Flac 1.1.2 -8 Results
314,591,233 A Perfect Circle - Mer De Noms.flac

Flac 1.1.2_CVS -8 Results
311,793,269 tukey(0.875).flac
311,795,907 hann.flac
311,799,691 tukey(0.75).flac
311,800,567 connes.flac
311,800,932 blackman.flac
311,805,213 kaiser_bessel.flac
311,807,392 tukey(0.625).flac
311,819,838 nuttall.flac
311,820,794 triangle.flac
311,821,621 blackman_harris_4term_92db.flac
311,824,423 tukey(0.5).flac
311,836,389 hamming.flac
311,839,182 welch.flac
311,842,671 gauss(0.25).flac
311,911,330 flattop.flac
311,911,482 tukey(0.25).flac
311,973,386 tukey(0.125).flac
312,003,993 gauss(0.5).flac
312,011,598 tukey(0.0625).flac
312,028,774 gauss(0.125).flac
312,029,537 tukey(0.03125).flac
312,050,233 tukey(0.015625).flac
312,059,872 tukey(0.0078125).flac
312,319,289 gauss(0.0625).flac
312,783,031 gauss(0.03125).flac
313,596,101 gauss(0.015625).flac
314,287,375 bartlett_hann.flac
314,591,231 rectangle.flac
314,927,922 gauss(0.0078125).flac
314,978,062 bartlett.flac

Synthetic Soul

May 4 2006, 02:12

QUOTE(Cartman_Sr @ May 3 2006, 10:01 PM)

Need any more tests?

I hope so as I've got 1,680 files being encoded at home as I type!

I intend to post results on 28 files testing tukey values from 0.1 to 1 in 0.1 increments for compression settings 3 through 8.

I think tukey looks like an interesting candidate at the moment so I wanted to try to pin it down a little more.

Afterward I may look at some of the other top contenders, like welch, connes, hamming, etc., if I have time.

Edit: Updated figures (I thought I was doing tukey(0) but I'm not)

A_Man_Eating_Duck

May 4 2006, 02:21

QUOTE(Synthetic Soul @ May 4 2006, 08:12 PM)

I hope so as I've got 1,848 files being encoded at home as I type!

How are you doing that many at one time? I would like to encode a few full albums over night to help with the comparison

Synthetic Soul

May 4 2006, 03:05

Good old batch files again.

I have a folder with 7 subfolders: 3; 4; 5; 6; 7; 8; source. "source" contains the source WAV files (with tagged FLAC counterparts so I can remember what they are!). Folders 3-8 will store the files for the given compression rate. Each of these folders has a batch file in it called "encode.bat". These are all exactly the same except for the compression level in the command line. "encode.bat" has a nested loop so that it loops through each tukey value and then through each wave file and creates a FLAC command line (10*28 combinations). It also writes the encoded file size to a text file (actually two: one per file and one per tukey value, as I'm not sure which one will be most use! (see disclaimer below)). Another batch file in "source" records all wave sizes to a text file. I then have "encodeall.bat" that just calls each "encode.bat" in turn.

The idea is that I should have, after the process, a file with the wave file sizes, and God knows how many text files with the compressed file sizes listed. I haven't yet worked out what I intend to do with those values, but I don't think I will have time to present it like my Yalac results. With that much data I'm going to have to do something though, or it will be useless.

NB: I created the system in a hurry last night. It could be more efficient and run from one batch file (three nested loops (tukey value/wave file/compression setting)). It could also delete the encoded file after recording its filesize - something that I only thought of on my cycle in!

I'm more than happy to zip up my folder system with batch files, but as I say, it's not as efficient as it could be, and it is also untested. I may get home tonight and find out that I am missing some data that makes the run useless!

Egor

May 4 2006, 03:34

Here is another script (I also updated the previuos) to test multiple -A functions (two):

CODE

for %%m in ("tukey(0.875)" "tukey(0.75)" "tukey(0.625)" "tukey(0.5)" "tukey(0.375)" "tukey(0.25)" "tukey(0.125)" "gauss(0.5)" "gauss(0.4375)" "gauss(0.375)" "gauss(0.25)" "gauss(0.1875)" "gauss(0.125)" "gauss(0.0625)" "hamming" "hann" "blackman" "kaiser_bessel" "connes" "triangle" "welch" "blackman_harris_4term_92db" "nuttall" "flattop" "bartlett_hann" "rectangle" "bartlett") do (for %%n in ("tukey(0.875)" "tukey(0.75)" "tukey(0.625)" "tukey(0.5)" "tukey(0.375)" "tukey(0.25)" "tukey(0.125)" "gauss(0.5)" "gauss(0.4375)" "gauss(0.375)" "gauss(0.25)" "gauss(0.1875)" "gauss(0.125)" "gauss(0.0625)" "hamming" "hann" "blackman" "kaiser_bessel" "connes" "triangle" "welch" "blackman_harris_4term_92db" "nuttall" "flattop" "bartlett_hann" "rectangle" "bartlett") do (start "%%n" /B /LOW /WAIT "flac.exe" -8 -A %%n -A %%m "test.wav" -o "%%n+%%m.flac" & dir /N "*.flac" >>"temp.txt" & del "*.flac"))
dir /N *.wav | find ".wav" >"report.txt"
type temp.txt | find ".flac" | sort >>"report.txt"

and the one with a colon as decimal point:

CODE

for %%m in ("tukey(0,875)" "tukey(0,75)" "tukey(0,625)" "tukey(0,5)" "tukey(0,375)" "tukey(0,25)" "tukey(0,125)" "gauss(0,5)" "gauss(0,4375)" "gauss(0,375)" "gauss(0,25)" "gauss(0,1875)" "gauss(0,125)" "gauss(0,0625)" "hamming" "hann" "blackman" "kaiser_bessel" "connes" "triangle" "welch" "blackman_harris_4term_92db" "nuttall" "flattop" "bartlett_hann" "rectangle" "bartlett") do (for %%n in ("tukey(0,875)" "tukey(0,75)" "tukey(0,625)" "tukey(0,5)" "tukey(0,375)" "tukey(0,25)" "tukey(0,125)" "gauss(0,5)" "gauss(0,4375)" "gauss(0,375)" "gauss(0,25)" "gauss(0,1875)" "gauss(0,125)" "gauss(0,0625)" "hamming" "hann" "blackman" "kaiser_bessel" "connes" "triangle" "welch" "blackman_harris_4term_92db" "nuttall" "flattop" "bartlett_hann" "rectangle" "bartlett") do (start "%%n" /B /LOW /WAIT "flac.exe" -8 -A %%n -A %%m "test.wav" -o "%%n+%%m.flac" & dir /N "*.flac" >>"temp.txt" & del "*.flac"))
dir /N *.wav | find ".wav" >"report.txt"
type temp.txt | find ".flac" | sort >>"report.txt"

And another report for CD album image:

CODE

Chris Botti - To love again
Jazz
flac -8

632 361 116 test.wav

293 452 753 tukey(0,5)
293 466 619 tukey(0,375)
293 471 878 tukey(0,625)
293 479 790 tukey(0,75)
293 490 410 tukey(0,25)
293 516 674 tukey(0,875)
293 520 670 tukey(0,125)
293 559 427 welch
293 562 552 hann
293 562 707 connes
293 562 834 gauss(0,4375)
293 573 929 hamming
293 579 661 gauss(0,375)
293 583 924 triangle
293 586 581 kaiser_bessel
293 586 902 blackman
293 595 904 nuttall
293 602 386 blackman_harris_4term_92db
293 624 948 gauss(0,25)
293 646 648 gauss(0,1875)
293 706 377 gauss(0,125)
293 710 505 flattop
293 796 740 gauss(0,5)
293 899 042 gauss(0,0625)
297 079 058 bartlett_hann
297 798 737 bartlett
297 827 546 rectangle

Edit. Added "flac -8" to the report.

Edit 2. 729 files is too much! So here is updated .cmd file. The previous one is also optimized.

A_Man_Eating_Duck

May 4 2006, 04:30

More results

Ben Folds - Ben Folds Live (full Album)
Style: Live Acoustic (piano, vocals)

CODE

Original Wav
754,620,428 Ben Folds - Ben Folds Live.wav

Flac 1.1.2 -8 Results
430,260,103 Ben Folds - Ben Folds Live.flac

Flac 1.1.2_CVS -8 Results
422,167,580 tukey(0.25).flac
422,177,455 tukey(0.125).flac
422,178,770 tukey(0.5).flac
422,187,253 tukey(0.0625).flac
422,189,073 welch.flac
422,191,020 tukey(0.625).flac
422,197,341 tukey(0.03125).flac
422,202,137 tukey(0.015625).flac
422,209,857 tukey(0.0078125).flac
422,215,337 tukey(0.75).flac
422,238,399 hamming.flac
422,245,879 tukey(0.875).flac
422,246,340 triangle.flac
422,252,225 connes.flac
422,276,057 hann.flac
422,336,776 blackman.flac
422,349,684 kaiser_bessel.flac
422,382,808 nuttall.flac
422,384,101 blackman_harris_4term_92db.flac
422,445,192 gauss(0.25).flac
422,549,584 flattop.flac
422,618,630 gauss(0.5).flac
422,674,069 gauss(0.125).flac
423,011,446 gauss(0.0625).flac
423,625,513 gauss(0.03125).flac
424,739,817 gauss(0.015625).flac
426,666,653 gauss(0.0078125).flac
429,013,707 bartlett_hann.flac
430,260,101 rectangle.flac
430,551,162 bartlett.flac

Synthetic Soul

May 4 2006, 05:21

Josh, do you have access to Excel? If so, what version?

FYI: Here's a spreadsheet (Excel 2003) with results for compression levels 3 and 4: flac-tukey-results.xls

If there's any point in using Excel I'll upload it again once all results are in. I can save as earlier versions I believe.

NB:

I have sorted the table by compression rate, which currently appears to be demonstrating that lower tukey values are best for higher compression (35-50%), and possibly very low compression (>80%), while higher values are better for everything inbetween (pop/rock). tukey(1) gets the best compression often, but the worst often also. It's early days yet though.
Best compression values are in green, worst in red.
The "File Info" sheet lists the file artist and title, otherwise files are referred to by their filename: "0" - "27".

Egor

May 4 2006, 05:34

QUOTE(Synthetic Soul @ May 4 2006, 06:21 PM)

Josh, do you have access to Excel? If so, what version?

BTW, Microsoft offers freeware Excel Viewer 2003 (for Windows, of course).
Also see Microsoft Office Converters and Viewers.

cyborg

May 4 2006, 05:47

RMB - Spring [Vocal Mix]

Genre: Trance

original size 66,342,908 bytes

length 6:16.093

CODE

45,222,877  tukey(0,25).flac
45,223,487  tukey(0,125).flac
45,224,288  tukey(0,0625).flac
45,224,613  tukey(0,03125).flac
45,224,795  tukey(0,015625).flac
45,225,023  tukey(0,5).flac
45,225,080  tukey(0,0078125).flac
45,226,586  welch.flac
45,227,352  tukey(0,625).flac
45,229,194  gauss(0,5).flac
45,229,582  tukey(0,75).flac
45,231,444  triangle.flac
45,232,047  tukey(0,875).flac
45,232,165  hamming.flac
45,232,256  connes.flac
45,234,394  hann.flac
45,239,975  blackman.flac
45,241,460  kaiser_bessel.flac
45,244,656  nuttall.flac
45,245,080  blackman_harris_4term_92db.flac
45,249,786  gauss(0,25).flac
45,253,224  rectangle.flac
45,258,785  flattop.flac
45,264,181  bartlett_hann.flac
45,275,458  gauss(0,125).flac
45,293,237  bartlett.flac
45,304,472  gauss(0,0625).flac
45,345,068  gauss(0,03125).flac
45,408,302  gauss(0,015625).flac
45,502,217  gauss(0,0078125).flac

Rollerball - Albinoni [Above & Beyond Remix]

Genre: Trance

original size 71,329,148 bytes

length 6:44.360

CODE

52,062,351  tukey(0,125).flac
52,062,482  tukey(0,25).flac
52,062,936  tukey(0,0625).flac
52,063,001  tukey(0,03125).flac
52,064,246  tukey(0,015625).flac
52,064,870  tukey(0,5).flac
52,067,221  tukey(0,0078125).flac
52,067,294  welch.flac
52,068,714  tukey(0,625).flac
52,070,781  tukey(0,75).flac
52,072,013  triangle.flac
52,074,425  tukey(0,875).flac
52,074,837  connes.flac
52,075,728  hann.flac
52,076,803  hamming.flac
52,077,693  gauss(0,5).flac
52,083,581  blackman.flac
52,086,202  kaiser_bessel.flac
52,088,943  blackman_harris_4term_92db.flac
52,089,947  nuttall.flac
52,094,941  gauss(0,25).flac
52,108,117  flattop.flac
52,125,576  gauss(0,125).flac
52,132,985  rectangle.flac
52,145,054  bartlett_hann.flac
52,161,993  gauss(0,0625).flac
52,188,425  bartlett.flac
52,220,056  gauss(0,03125).flac
52,312,455  gauss(0,015625).flac
52,455,149  gauss(0,0078125).flac

Egor

May 4 2006, 08:18

Multiple -A parameters produce smaller sizes, report:

CODE

Michael Buble - A foggy day (in London town)
Jazz Vocal / Popular
flac -8

26 730 524 test.wav

17 942 993 gauss(0,1875)+tukey(0,125).flac
17 942 994 tukey(0,125)+gauss(0,1875).flac
17 943 062 gauss(0,125)+tukey(0,125).flac
17 943 063 tukey(0,125)+gauss(0,125).flac
17 943 254 tukey(0,125)+flattop.flac
17 943 256 flattop+tukey(0,125).flac
17 943 671 gauss(0,0625)+tukey(0,125).flac
17 943 671 tukey(0,125)+gauss(0,0625).flac
17 944 064 tukey(0,25)+gauss(0,125).flac
17 944 065 gauss(0,125)+tukey(0,25).flac
17 944 135 tukey(0,125)+gauss(0,25).flac
17 944 136 gauss(0,25)+tukey(0,125).flac
17 944 164 tukey(0,375)+gauss(0,125).flac
17 944 165 gauss(0,125)+tukey(0,375).flac
17 944 258 gauss(0,1875)+tukey(0,25).flac
17 944 260 tukey(0,25)+gauss(0,1875).flac
17 944 326 gauss(0,0625)+tukey(0,375).flac
17 944 328 tukey(0,375)+gauss(0,0625).flac
17 944 355 flattop+tukey(0,25).flac
17 944 363 tukey(0,25)+flattop.flac
17 944 533 tukey(0,25)+gauss(0,0625).flac
17 944 534 gauss(0,0625)+tukey(0,25).flac
17 944 738 tukey(0,125)+blackman_harris_4term_92db.flac
17 944 739 blackman_harris_4term_92db+tukey(0,125).flac
17 944 851 nuttall+tukey(0,125).flac
17 944 854 tukey(0,125)+nuttall.flac
17 944 941 gauss(0,1875)+tukey(0,375).flac
17 944 944 tukey(0,375)+gauss(0,1875).flac
17 944 982 tukey(0,375)+flattop.flac
17 944 984 flattop+tukey(0,375).flac
17 945 339 gauss(0,25)+tukey(0,25).flac
17 945 339 kaiser_bessel+tukey(0,125).flac
17 945 339 tukey(0,25)+gauss(0,25).flac
17 945 340 tukey(0,125)+kaiser_bessel.flac
17 945 848 blackman_harris_4term_92db+tukey(0,25).flac
17 945 848 tukey(0,25)+blackman_harris_4term_92db.flac
17 945 864 gauss(0,0625)+tukey(0,5).flac
17 945 864 gauss(0,125)+tukey(0,5).flac
17 945 866 tukey(0,5)+gauss(0,125).flac
17 945 867 tukey(0,5)+gauss(0,0625).flac
17 946 033 blackman+tukey(0,125).flac
17 946 033 tukey(0,125)+blackman.flac
17 946 131 tukey(0,375)+gauss(0,25).flac
17 946 134 gauss(0,25)+tukey(0,375).flac
17 946 607 tukey(0,25)+nuttall.flac
17 946 611 nuttall+tukey(0,25).flac
17 946 684 gauss(0,1875)+tukey(0,5).flac
17 946 684 tukey(0,5)+gauss(0,1875).flac
17 946 693 tukey(0,25)+kaiser_bessel.flac
17 946 704 kaiser_bessel+tukey(0,25).flac
17 946 814 blackman_harris_4term_92db+tukey(0,375).flac
17 946 815 tukey(0,375)+blackman_harris_4term_92db.flac
17 946 818 tukey(0,625)+gauss(0,0625).flac
17 946 821 gauss(0,0625)+tukey(0,625).flac
17 946 880 tukey(0,5)+flattop.flac
17 946 886 flattop+tukey(0,5).flac
17 947 153 tukey(0,125)+gauss(0,375).flac
17 947 155 gauss(0,375)+tukey(0,125).flac
17 947 288 gauss(0,125)+tukey(0,625).flac
17 947 289 tukey(0,625)+gauss(0,125).flac
17 947 370 tukey(0,25)+blackman.flac
17 947 373 blackman+tukey(0,25).flac
17 947 413 tukey(0,375)+nuttall.flac
17 947 414 nuttall+tukey(0,375).flac
17 947 626 hann+tukey(0,125).flac
17 947 630 tukey(0,125)+hann.flac
17 947 735 tukey(0,375)+kaiser_bessel.flac
17 947 745 kaiser_bessel+tukey(0,375).flac
17 947 815 gauss(0,0625)+welch.flac
17 947 817 welch+gauss(0,0625).flac
17 948 156 tukey(0,625)+gauss(0,1875).flac
17 948 157 gauss(0,1875)+tukey(0,625).flac
17 948 276 gauss(0,0625)+tukey(0,75).flac
17 948 276 tukey(0,75)+gauss(0,0625).flac
17 948 372 gauss(0,125)+welch.flac
17 948 372 welch+gauss(0,125).flac
17 948 377 blackman+tukey(0,375).flac
17 948 378 tukey(0,375)+blackman.flac
17 948 440 tukey(0,5)+gauss(0,25).flac
17 948 443 gauss(0,25)+tukey(0,5).flac
17 948 641 tukey(0,875)+tukey(0,125).flac
17 948 643 tukey(0,125)+tukey(0,875).flac
17 948 769 flattop+tukey(0,625).flac
17 948 773 tukey(0,25)+gauss(0,375).flac
17 948 774 tukey(0,625)+flattop.flac
17 948 776 gauss(0,375)+tukey(0,25).flac
17 948 802 gauss(0,125)+tukey(0,75).flac
17 948 807 tukey(0,75)+gauss(0,125).flac
17 948 957 connes+tukey(0,125).flac
17 948 957 tukey(0,125)+connes.flac
17 949 133 blackman_harris_4term_92db+tukey(0,5).flac
17 949 133 tukey(0,5)+blackman_harris_4term_92db.flac
17 949 218 tukey(0,125)+gauss(0,4375).flac
17 949 219 gauss(0,4375)+tukey(0,125).flac
17 949 351 hann+tukey(0,25).flac
17 949 353 tukey(0,25)+hann.flac
17 949 559 hamming+tukey(0,125).flac
17 949 561 tukey(0,125)+hamming.flac
17 949 610 gauss(0,1875)+welch.flac
17 949 619 welch+gauss(0,1875).flac
17 949 731 tukey(0,75)+tukey(0,125).flac
17 949 733 tukey(0,125)+tukey(0,75).flac
17 949 958 tukey(0,125)+triangle.flac
17 949 964 triangle+tukey(0,125).flac
17 949 964 tukey(0,5)+nuttall.flac
17 949 965 nuttall+tukey(0,5).flac
17 950 046 welch+flattop.flac
17 950 050 flattop+welch.flac
17 950 090 gauss(0,375)+tukey(0,375).flac
17 950 090 tukey(0,375)+gauss(0,375).flac
17 950 196 gauss(0,25)+tukey(0,625).flac
17 950 202 tukey(0,625)+gauss(0,25).flac
17 950 420 tukey(0,75)+gauss(0,1875).flac
17 950 426 gauss(0,1875)+tukey(0,75).flac
17 950 589 connes+tukey(0,25).flac
17 950 590 tukey(0,25)+connes.flac
17 950 615 flattop+tukey(0,75).flac
17 950 618 tukey(0,75)+flattop.flac
17 950 630 gauss(0,0625)+gauss(0,5).flac
17 950 632 gauss(0,5)+gauss(0,0625).flac
17 950 650 tukey(0,25)+tukey(0,875).flac
17 950 652 tukey(0,875)+tukey(0,25).flac
17 950 751 tukey(0,5)+kaiser_bessel.flac
17 950 756 kaiser_bessel+tukey(0,5).flac
17 950 774 tukey(0,875)+gauss(0,0625).flac
17 950 775 gauss(0,0625)+tukey(0,875).flac
17 950 927 hamming+gauss(0,0625).flac
17 950 929 gauss(0,0625)+hamming.flac
17 950 948 gauss(0,4375)+gauss(0,0625).flac
17 950 950 gauss(0,0625)+gauss(0,4375).flac
17 951 035 hamming+tukey(0,25).flac
17 951 041 tukey(0,25)+hamming.flac
17 951 042 hann+tukey(0,375).flac
17 951 044 tukey(0,375)+hann.flac
17 951 058 gauss(0,5)+gauss(0,125).flac
17 951 059 gauss(0,125)+gauss(0,5).flac
17 951 129 tukey(0,25)+gauss(0,4375).flac
17 951 130 gauss(0,4375)+tukey(0,25).flac
17 951 342 tukey(0,5)+blackman.flac
17 951 345 blackman+tukey(0,5).flac
17 951 377 gauss(0,0625)+triangle.flac
17 951 378 triangle+gauss(0,0625).flac
17 951 465 gauss(0,0625)+connes.flac
17 951 466 connes+gauss(0,0625).flac
17 951 532 tukey(0,75)+tukey(0,25).flac
17 951 536 tukey(0,25)+tukey(0,75).flac
17 951 563 tukey(0,625)+blackman_harris_4term_92db.flac
17 951 574 blackman_harris_4term_92db+tukey(0,625).flac
17 951 579 tukey(0,25)+triangle.flac
17 951 581 triangle+tukey(0,25).flac
17 951 656 tukey(0,875)+gauss(0,125).flac
17 951 658 gauss(0,125)+tukey(0,875).flac
17 951 690 gauss(0,25)+welch.flac
17 951 691 welch+gauss(0,25).flac
17 951 931 tukey(0,625)+tukey(0,125).flac
17 951 935 tukey(0,125)+tukey(0,625).flac
17 952 019 gauss(0,125)+triangle.flac
17 952 019 triangle+gauss(0,125).flac
17 952 054 connes+tukey(0,375).flac
17 952 055 tukey(0,375)+connes.flac
17 952 066 hamming+gauss(0,125).flac
17 952 067 gauss(0,125)+hamming.flac
17 952 081 tukey(0,375)+tukey(0,875).flac
17 952 081 tukey(0,875)+tukey(0,375).flac
17 952 139 nuttall+tukey(0,625).flac
17 952 139 tukey(0,625)+nuttall.flac
17 952 284 tukey(0,125)+gauss(0,5).flac
17 952 286 gauss(0,5)+tukey(0,125).flac
17 952 314 gauss(0,4375)+gauss(0,125).flac
17 952 319 gauss(0,125)+gauss(0,4375).flac
17 952 392 gauss(0,0625)+hann.flac
17 952 394 hann+gauss(0,0625).flac
17 952 403 blackman_harris_4term_92db+welch.flac
17 952 403 welch+blackman_harris_4term_92db.flac
17 952 513 connes+gauss(0,125).flac
17 952 518 gauss(0,125)+connes.flac
17 952 541 gauss(0,5)+gauss(0,1875).flac
17 952 546 gauss(0,1875)+gauss(0,5).flac
17 952 693 gauss(0,4375)+tukey(0,375).flac
17 952 693 tukey(0,375)+gauss(0,4375).flac
17 952 863 gauss(0,5)+flattop.flac
17 952 867 flattop+gauss(0,5).flac
17 952 927 tukey(0,375)+hamming.flac
17 952 930 hamming+tukey(0,375).flac
17 952 931 tukey(0,625)+kaiser_bessel.flac
17 952 934 kaiser_bessel+tukey(0,625).flac
17 953 025 gauss(0,25)+tukey(0,75).flac
17 953 026 tukey(0,75)+gauss(0,25).flac
17 953 061 tukey(0,5)+gauss(0,375).flac
17 953 062 gauss(0,375)+tukey(0,5).flac
17 953 152 gauss(0,1875)+tukey(0,875).flac
17 953 154 tukey(0,875)+gauss(0,1875).flac
17 953 292 tukey(0,375)+triangle.flac
17 953 294 triangle+tukey(0,375).flac
17 953 326 welch+nuttall.flac
17 953 332 nuttall+welch.flac
17 953 335 gauss(0,1875)+triangle.flac
17 953 337 triangle+gauss(0,1875).flac
17 953 497 gauss(0,125)+hann.flac
17 953 499 hann+gauss(0,125).flac
17 953 524 gauss(0,4375)+gauss(0,1875).flac
17 953 525 gauss(0,1875)+gauss(0,4375).flac
17 953 585 tukey(0,375)+tukey(0,75).flac
17 953 585 tukey(0,75)+tukey(0,375).flac
17 953 614 hamming+gauss(0,1875).flac
17 953 616 gauss(0,1875)+hamming.flac
17 953 639 tukey(0,625)+blackman.flac
17 953 640 blackman+tukey(0,625).flac
17 953 718 tukey(0,125)+welch.flac
17 953 719 welch+tukey(0,125).flac
17 953 769 hann+tukey(0,5).flac
17 953 770 tukey(0,5)+hann.flac
17 953 840 gauss(0,1875)+connes.flac
17 953 841 connes+gauss(0,1875).flac
17 953 850 flattop+tukey(0,875).flac
17 953 850 tukey(0,875)+flattop.flac
17 953 999 gauss(0,375)+gauss(0,0625).flac
17 954 000 gauss(0,0625)+gauss(0,375).flac
17 954 004 tukey(0,75)+blackman_harris_4term_92db.flac
17 954 007 blackman_harris_4term_92db+tukey(0,75).flac
17 954 011 tukey(0,25)+tukey(0,625).flac
17 954 013 tukey(0,625)+tukey(0,25).flac
17 954 097 bartlett+tukey(0,75).flac
17 954 097 tukey(0,75)+bartlett.flac
17 954 265 tukey(0,5)+tukey(0,125).flac
17 954 269 tukey(0,125)+tukey(0,5).flac
17 954 284 kaiser_bessel+welch.flac
17 954 287 welch+kaiser_bessel.flac
17 954 305 flattop+gauss(0,4375).flac
17 954 309 gauss(0,4375)+flattop.flac
17 954 348 bartlett+tukey(0,625).flac
17 954 348 tukey(0,625)+bartlett.flac
17 954 354 triangle+flattop.flac
17 954 356 flattop+triangle.flac
17 954 430 welch+blackman.flac
17 954 437 blackman+welch.flac
17 954 495 flattop+hamming.flac
17 954 500 hamming+flattop.flac
17 954 532 tukey(0,25)+gauss(0,5).flac
17 954 533 gauss(0,5)+tukey(0,25).flac
17 954 648 connes+flattop.flac
17 954 648 flattop+connes.flac
17 954 699 gauss(0,5)+gauss(0,25).flac
17 954 700 gauss(0,25)+gauss(0,5).flac
17 954 716 tukey(0,75)+nuttall.flac
17 954 718 nuttall+tukey(0,75).flac
17 954 803 tukey(0,5)+bartlett.flac
17 954 804 bartlett+tukey(0,5).flac
17 954 867 tukey(0,875)+bartlett.flac
17 954 874 bartlett+tukey(0,875).flac
17 954 925 tukey(0,5)+tukey(0,875).flac
17 954 927 tukey(0,875)+tukey(0,5).flac
17 954 998 bartlett+hann.flac
17 954 998 hann+bartlett.flac
17 955 141 bartlett+connes.flac
17 955 143 connes+bartlett.flac
17 955 170 gauss(0,4375)+bartlett.flac
17 955 172 bartlett+gauss(0,4375).flac
17 955 328 tukey(0,5)+connes.flac
17 955 335 connes+tukey(0,5).flac
17 955 343 bartlett+hamming.flac
17 955 347 hamming+bartlett.flac
17 955 358 kaiser_bessel+bartlett.flac
17 955 359 bartlett+kaiser_bessel.flac
17 955 369 gauss(0,1875)+hann.flac
17 955 371 hann+gauss(0,1875).flac
17 955 384 gauss(0,125)+gauss(0,375).flac
17 955 389 gauss(0,375)+gauss(0,125).flac
17 955 447 gauss(0,4375)+tukey(0,5).flac
17 955 449 tukey(0,5)+gauss(0,4375).flac
17 955 520 gauss(0,5)+tukey(0,375).flac
17 955 522 tukey(0,375)+gauss(0,5).flac
17 955 525 blackman+bartlett.flac
17 955 526 bartlett+blackman.flac
17 955 560 tukey(0,5)+hamming.flac
17 955 567 hamming+tukey(0,5).flac
17 955 595 blackman_harris_4term_92db+bartlett.flac
17 955 596 bartlett+blackman_harris_4term_92db.flac
17 955 602 bartlett+tukey(0,375).flac
17 955 609 tukey(0,375)+bartlett.flac
17 955 648 tukey(0,75)+kaiser_bessel.flac
17 955 653 kaiser_bessel+tukey(0,75).flac
17 955 657 gauss(0,375)+bartlett.flac
17 955 658 bartlett+gauss(0,375).flac
17 955 666 blackman_harris_4term_92db+gauss(0,5).flac
17 955 666 gauss(0,5)+blackman_harris_4term_92db.flac
17 955 672 gauss(0,375)+tukey(0,625).flac
17 955 675 tukey(0,625)+gauss(0,375).flac
17 955 728 tukey(0,875)+gauss(0,25).flac
17 955 734 gauss(0,25)+tukey(0,875).flac
17 955 860 welch+tukey(0,25).flac
17 955 863 tukey(0,25)+welch.flac
17 955 889 welch+bartlett.flac
17 955 890 bartlett+welch.flac
17 955 964 triangle+bartlett.flac
17 955 967 bartlett+triangle.flac
17 956 061 tukey(0,5)+triangle.flac
17 956 068 triangle+tukey(0,5).flac
17 956 070 tukey(0,375)+tukey(0,625).flac
17 956 075 tukey(0,625)+tukey(0,375).flac
17 956 134 triangle+gauss(0,25).flac
17 956 135 gauss(0,25)+triangle.flac
17 956 197 hann+flattop.flac
17 956 199 flattop+hann.flac
17 956 217 gauss(0,4375)+gauss(0,25).flac
17 956 223 gauss(0,25)+gauss(0,4375).flac
17 956 265 blackman+gauss(0,0625).flac
17 956 266 gauss(0,0625)+blackman.flac
17 956 411 gauss(0,375)+welch.flac
17 956 415 welch+gauss(0,375).flac
17 956 423 tukey(0,75)+blackman.flac
17 956 424 blackman+tukey(0,75).flac
17 956 470 gauss(0,25)+hamming.flac
17 956 471 hamming+gauss(0,25).flac
17 956 504 gauss(0,5)+nuttall.flac
17 956 506 nuttall+gauss(0,5).flac
17 956 516 bartlett+gauss(0,25).flac
17 956 517 gauss(0,25)+bartlett.flac
17 956 614 tukey(0,5)+tukey(0,75).flac
17 956 618 tukey(0,75)+tukey(0,5).flac
17 956 670 connes+gauss(0,25).flac
17 956 670 gauss(0,25)+connes.flac
17 956 777 tukey(0,625)+hann.flac
17 956 780 hann+tukey(0,625).flac
17 956 833 tukey(0,25)+tukey(0,5).flac
17 956 834 tukey(0,5)+tukey(0,25).flac
17 956 905 rectangle+blackman_harris_4term_92db.flac
17 956 906 blackman_harris_4term_92db+rectangle.flac
17 956 941 gauss(0,5)+kaiser_bessel.flac
17 956 941 kaiser_bessel+gauss(0,5).flac
17 957 060 rectangle+kaiser_bessel.flac
17 957 061 kaiser_bessel+rectangle.flac
17 957 112 gauss(0,0625)+kaiser_bessel.flac
17 957 113 kaiser_bessel+gauss(0,0625).flac
17 957 147 rectangle+gauss(0,25).flac
17 957 151 gauss(0,25)+rectangle.flac
17 957 180 tukey(0,375)+tukey(0,125).flac
17 957 181 bartlett+gauss(0,1875).flac
17 957 181 tukey(0,125)+tukey(0,375).flac
17 957 182 gauss(0,1875)+bartlett.flac
17 957 183 nuttall+bartlett.flac
17 957 184 bartlett+nuttall.flac
17 957 184 welch+hann.flac
17 957 184 welch+tukey(0,375).flac
17 957 187 hann+welch.flac
17 957 189 tukey(0,375)+welch.flac
17 957 210 blackman+rectangle.flac
17 957 211 rectangle+blackman.flac
17 957 224 gauss(0,1875)+gauss(0,375).flac
17 957 226 blackman_harris_4term_92db+tukey(0,875).flac
17 957 229 gauss(0,375)+gauss(0,1875).flac
17 957 232 tukey(0,875)+blackman_harris_4term_92db.flac
17 957 392 triangle+blackman_harris_4term_92db.flac
17 957 395 blackman_harris_4term_92db+triangle.flac
17 957 490 rectangle+gauss(0,1875).flac
17 957 495 gauss(0,1875)+rectangle.flac
17 957 575 blackman_harris_4term_92db+hamming.flac
17 957 576 hamming+blackman_harris_4term_92db.flac
17 957 579 rectangle+gauss(0,375).flac
17 957 583 gauss(0,375)+rectangle.flac
17 957 597 bartlett+tukey(0,25).flac
17 957 597 tukey(0,25)+bartlett.flac
17 957 627 gauss(0,4375)+blackman_harris_4term_92db.flac
17 957 631 blackman_harris_4term_92db+gauss(0,4375).flac
17 957 686 blackman+gauss(0,125).flac
17 957 687 gauss(0,125)+blackman.flac
17 957 711 tukey(0,75)+rectangle.flac
17 957 713 rectangle+tukey(0,75).flac
17 957 723 tukey(0,625)+gauss(0,4375).flac
17 957 724 gauss(0,4375)+tukey(0,625).flac
17 957 734 blackman+gauss(0,5).flac
17 957 735 gauss(0,5)+blackman.flac
17 957 739 connes+tukey(0,625).flac
17 957 740 rectangle+hann.flac
17 957 743 tukey(0,625)+connes.flac
17 957 744 hann+rectangle.flac
17 957 801 bartlett+flattop.flac
17 957 801 flattop+bartlett.flac
17 957 846 tukey(0,625)+tukey(0,875).flac
17 957 852 tukey(0,875)+tukey(0,625).flac
17 957 897 flattop+gauss(0,375).flac
17 957 902 gauss(0,375)+flattop.flac
17 957 907 hann+gauss(0,25).flac
17 957 911 gauss(0,25)+hann.flac
17 958 000 hamming+tukey(0,625).flac
17 958 006 tukey(0,625)+hamming.flac
17 958 025 rectangle+tukey(0,875).flac
17 958 027 tukey(0,875)+rectangle.flac
17 958 028 tukey(0,5)+gauss(0,5).flac
17 958 030 gauss(0,5)+tukey(0,5).flac
17 958 057 nuttall+gauss(0,4375).flac
17 958 065 gauss(0,4375)+nuttall.flac
17 958 078 tukey(0,875)+nuttall.flac
17 958 080 nuttall+tukey(0,875).flac
17 958 108 rectangle+gauss(0,4375).flac
17 958 113 gauss(0,4375)+rectangle.flac
17 958 118 blackman_harris_4term_92db+connes.flac
17 958 123 connes+blackman_harris_4term_92db.flac
17 958 153 nuttall+hamming.flac
17 958 155 hamming+nuttall.flac
17 958 273 tukey(0,625)+triangle.flac
17 958 275 tukey(0,75)+gauss(0,375).flac
17 958 278 gauss(0,375)+tukey(0,75).flac
17 958 279 triangle+tukey(0,625).flac
17 958 291 gauss(0,5)+bartlett.flac
17 958 292 bartlett+gauss(0,5).flac
17 958 308 rectangle+connes.flac
17 958 309 connes+rectangle.flac
17 958 331 connes+welch.flac
17 958 333 welch+connes.flac
17 958 342 rectangle+flattop.flac
17 958 347 flattop+rectangle.flac
17 958 375 rectangle+nuttall.flac
17 958 378 nuttall+rectangle.flac
17 958 442 tukey(0,875)+welch.flac
17 958 446 welch+tukey(0,875).flac
17 958 550 triangle+nuttall.flac
17 958 552 nuttall+triangle.flac
17 958 562 rectangle+hamming.flac
17 958 564 hamming+rectangle.flac
17 958 595 bartlett_hann+tukey(0,375).flac
17 958 597 tukey(0,375)+bartlett_hann.flac
17 958 653 gauss(0,125)+kaiser_bessel.flac
17 958 656 kaiser_bessel+gauss(0,125).flac
17 958 681 bartlett+tukey(0,125).flac
17 958 684 tukey(0,125)+bartlett.flac
17 958 773 gauss(0,4375)+welch.flac
17 958 775 welch+gauss(0,4375).flac
17 958 778 triangle+kaiser_bessel.flac
17 958 782 welch+hamming.flac
17 958 785 kaiser_bessel+triangle.flac
17 958 789 hamming+welch.flac
17 958 843 connes+nuttall.flac
17 958 843 nuttall+connes.flac
17 958 976 tukey(0,5)+tukey(0,375).flac
17 958 977 tukey(0,375)+tukey(0,5).flac
17 959 015 bartlett_hann+tukey(0,5).flac
17 959 016 tukey(0,5)+bartlett_hann.flac
17 959 037 tukey(0,875)+kaiser_bessel.flac
17 959 043 kaiser_bessel+tukey(0,875).flac
17 959 112 rectangle+tukey(0,625).flac
17 959 113 tukey(0,625)+rectangle.flac
17 959 150 tukey(0,75)+hann.flac
17 959 165 tukey(0,625)+tukey(0,5).flac
17 959 168 hann+tukey(0,75).flac
17 959 168 tukey(0,5)+tukey(0,625).flac
17 959 176 hamming+kaiser_bessel.flac
17 959 178 kaiser_bessel+hamming.flac
17 959 189 triangle+welch.flac
17 959 192 welch+triangle.flac
17 959 272 blackman_harris_4term_92db+hann.flac
17 959 272 hann+blackman_harris_4term_92db.flac
17 959 273 gauss(0,375)+gauss(0,5).flac
17 959 277 gauss(0,5)+gauss(0,375).flac
17 959 321 gauss(0,4375)+kaiser_bessel.flac
17 959 325 kaiser_bessel+gauss(0,4375).flac
17 959 343 rectangle+triangle.flac
17 959 344 triangle+rectangle.flac
17 959 369 tukey(0,75)+tukey(0,625).flac
17 959 372 tukey(0,625)+tukey(0,75).flac
17 959 383 welch+tukey(0,5).flac
17 959 387 tukey(0,5)+welch.flac
17 959 437 welch+tukey(0,75).flac
17 959 441 tukey(0,75)+welch.flac
17 959 509 blackman+gauss(0,1875).flac
17 959 509 gauss(0,1875)+blackman.flac
17 959 541 blackman_harris_4term_92db+gauss(0,0625).flac
17 959 545 gauss(0,0625)+blackman_harris_4term_92db.flac
17 959 564 tukey(0,625)+gauss(0,5).flac
17 959 567 gauss(0,5)+tukey(0,625).flac
17 959 625 triangle+blackman.flac
17 959 626 blackman+triangle.flac
17 959 824 blackman+gauss(0,4375).flac
17 959 824 gauss(0,4375)+blackman.flac
17 959 890 connes+kaiser_bessel.flac
17 959 890 tukey(0,875)+blackman.flac
17 959 891 kaiser_bessel+connes.flac
17 959 895 blackman+tukey(0,875).flac
17 959 908 tukey(0,75)+gauss(0,4375).flac
17 959 909 gauss(0,4375)+tukey(0,75).flac
17 959 915 tukey(0,625)+bartlett_hann.flac
17 959 917 bartlett_hann+tukey(0,625).flac
17 959 944 tukey(0,25)+tukey(0,375).flac
17 959 947 tukey(0,375)+tukey(0,25).flac
17 959 994 bartlett_hann+tukey(0,25).flac
17 959 994 tukey(0,25)+bartlett_hann.flac
17 960 093 gauss(0,25)+gauss(0,375).flac
17 960 093 gauss(0,375)+gauss(0,25).flac
17 960 121 hamming+blackman.flac
17 960 127 blackman+hamming.flac
17 960 141 rectangle+gauss(0,125).flac
17 960 142 gauss(0,125)+rectangle.flac
17 960 169 tukey(0,75)+triangle.flac
17 960 170 hann+nuttall.flac
17 960 170 triangle+tukey(0,75).flac
17 960 179 nuttall+hann.flac
17 960 201 gauss(0,125)+bartlett.flac
17 960 202 bartlett+gauss(0,125).flac
17 960 211 tukey(0,75)+hamming.flac
17 960 212 hamming+tukey(0,75).flac
17 960 269 nuttall+gauss(0,0625).flac
17 960 273 gauss(0,0625)+nuttall.flac
17 960 308 gauss(0,5)+hann.flac
17 960 313 hann+gauss(0,5).flac
17 960 323 tukey(0,75)+tukey(0,875).flac
17 960 335 rectangle+tukey(0,5).flac
17 960 335 tukey(0,5)+rectangle.flac
17 960 336 tukey(0,875)+tukey(0,75).flac
17 960 383 welch+tukey(0,625).flac
17 960 385 tukey(0,625)+welch.flac
17 960 485 bartlett_hann+tukey(0,125).flac
17 960 489 tukey(0,125)+bartlett_hann.flac
17 960 517 tukey(0,75)+gauss(0,5).flac
17 960 522 gauss(0,5)+tukey(0,75).flac
17 960 531 tukey(0,75)+connes.flac
17 960 538 connes+tukey(0,75).flac
17 960 592 bartlett_hann+tukey(0,75).flac
17 960 592 tukey(0,75)+bartlett_hann.flac
17 960 774 blackman+connes.flac
17 960 774 connes+blackman.flac
17 960 951 hann+kaiser_bessel.flac
17 960 954 kaiser_bessel+hann.flac
17 961 042 rectangle+welch.flac
17 961 043 gauss(0,1875)+kaiser_bessel.flac
17 961 043 welch+rectangle.flac
17 961 051 kaiser_bessel+gauss(0,1875).flac
17 961 109 flattop+blackman.flac
17 961 113 blackman+flattop.flac
17 961 182 tukey(0,875)+gauss(0,375).flac
17 961 183 gauss(0,375)+tukey(0,875).flac
17 961 235 hann+triangle.flac
17 961 241 triangle+hann.flac
17 961 247 gauss(0,5)+tukey(0,875).flac
17 961 250 tukey(0,875)+gauss(0,5).flac
17 961 290 triangle+gauss(0,375).flac
17 961 293 gauss(0,375)+triangle.flac
17 961 357 gauss(0,375)+blackman_harris_4term_92db.flac
17 961 358 blackman_harris_4term_92db+gauss(0,375).flac
17 961 457 blackman_harris_4term_92db+gauss(0,125).flac
17 961 462 gauss(0,125)+blackman_harris_4term_92db.flac
17 961 462 welch+gauss(0,5).flac
17 961 467 gauss(0,5)+welch.flac
17 961 561 gauss(0,5)+connes.flac
17 961 572 connes+gauss(0,5).flac
17 961 729 gauss(0,4375)+gauss(0,375).flac
17 961 735 gauss(0,375)+gauss(0,4375).flac
17 961 747 blackman+hann.flac
17 961 752 hann+blackman.flac
17 961 774 tukey(0,875)+gauss(0,4375).flac
17 961 782 gauss(0,4375)+tukey(0,875).flac
17 961 818 gauss(0,4375)+gauss(0,5).flac
17 961 825 gauss(0,5)+gauss(0,4375).flac
17 961 879 flattop+kaiser_bessel.flac
17 961 884 kaiser_bessel+flattop.flac
17 961 950 tukey(0,125)+tukey(0,25).flac
17 961 950 tukey(0,25)+tukey(0,125).flac
17 961 955 bartlett_hann+welch.flac
17 961 955 welch+bartlett_hann.flac
17 961 998 nuttall+gauss(0,125).flac
17 962 000 gauss(0,125)+nuttall.flac
17 962 014 triangle+tukey(0,875).flac
17 962 017 tukey(0,875)+triangle.flac
17 962 079 hamming+gauss(0,375).flac
17 962 081 gauss(0,375)+hamming.flac
17 962 111 nuttall+gauss(0,375).flac
17 962 112 gauss(0,375)+nuttall.flac
17 962 118 tukey(0,875)+hann.flac
17 962 120 hann+tukey(0,875).flac
17 962 279 connes+gauss(0,375).flac
17 962 282 gauss(0,375)+connes.flac
17 962 288 hamming+hann.flac
17 962 296 hann+hamming.flac
17 962 302 hamming+gauss(0,5).flac
17 962 305 gauss(0,5)+hamming.flac
17 962 322 gauss(0,4375)+hann.flac
17 962 330 hann+gauss(0,4375).flac
17 962 373 tukey(0,875)+bartlett_hann.flac
17 962 375 bartlett_hann+tukey(0,875).flac
17 962 405 bartlett_hann+blackman_harris_4term_92db.flac
17 962 405 triangle+gauss(0,5).flac
17 962 406 blackman_harris_4term_92db+bartlett_hann.flac
17 962 409 gauss(0,5)+rectangle.flac
17 962 409 gauss(0,5)+triangle.flac
17 962 409 rectangle+gauss(0,5).flac
17 962 449 bartlett_hann+gauss(0,1875).flac
17 962 449 gauss(0,1875)+bartlett_hann.flac
17 962 508 connes+triangle.flac
17 962 509 triangle+connes.flac
17 962 515 gauss(0,4375)+triangle.flac
17 962 516 triangle+gauss(0,4375).flac
17 962 529 tukey(0,375)+rectangle.flac
17 962 530 rectangle+tukey(0,375).flac
17 962 558 hamming+tukey(0,875).flac
17 962 565 tukey(0,875)+hamming.flac
17 962 716 gauss(0,25)+bartlett_hann.flac
17 962 717 bartlett_hann+gauss(0,25).flac
17 962 741 kaiser_bessel+bartlett_hann.flac
17 962 743 bartlett_hann+kaiser_bessel.flac
17 962 752 bartlett_hann+connes.flac
17 962 755 connes+bartlett_hann.flac
17 962 822 bartlett_hann+blackman.flac
17 962 822 blackman+bartlett_hann.flac
17 962 838 bartlett_hann+gauss(0,4375).flac
17 962 842 gauss(0,4375)+bartlett_hann.flac
17 962 845 bartlett_hann+hann.flac
17 962 849 hann+bartlett_hann.flac
17 962 898 gauss(0,375)+kaiser_bessel.flac
17 962 902 kaiser_bessel+gauss(0,375).flac
17 962 905 gauss(0,4375)+connes.flac
17 962 908 connes+gauss(0,4375).flac
17 963 021 bartlett_hann+flattop.flac
17 963 023 flattop+bartlett_hann.flac
17 963 040 hann+connes.flac
17 963 046 connes+hann.flac
17 963 062 gauss(0,0625)+gauss(0,25).flac
17 963 064 blackman+gauss(0,25).flac
17 963 064 gauss(0,25)+gauss(0,0625).flac
17 963 066 gauss(0,25)+blackman.flac
17 963 081 bartlett_hann+gauss(0,375).flac
17 963 084 gauss(0,375)+bartlett_hann.flac
17 963 106 connes+tukey(0,875).flac
17 963 109 tukey(0,875)+connes.flac
17 963 203 bartlett_hann+hamming.flac
17 963 204 hamming+bartlett_hann.flac
17 963 214 hamming+triangle.flac
17 963 217 triangle+hamming.flac
17 963 304 connes+hamming.flac
17 963 310 hamming+connes.flac
17 963 349 bartlett_hann+triangle.flac
17 963 350 triangle+bartlett_hann.flac
17 963 454 blackman+gauss(0,375).flac
17 963 455 gauss(0,375)+blackman.flac
17 963 486 hann+gauss(0,375).flac
17 963 487 gauss(0,375)+hann.flac
17 963 590 bartlett_hann+gauss(0,125).flac
17 963 595 gauss(0,125)+bartlett_hann.flac
17 963 630 gauss(0,1875)+blackman_harris_4term_92db.flac
17 963 633 blackman_harris_4term_92db+gauss(0,1875).flac
17 963 704 nuttall+bartlett_hann.flac
17 963 707 bartlett_hann+nuttall.flac
17 963 725 gauss(0,4375)+hamming.flac
17 963 726 hamming+gauss(0,4375).flac
17 963 931 gauss(0,25)+kaiser_bessel.flac
17 963 933 kaiser_bessel+gauss(0,25).flac
17 964 052 gauss(0,1875)+nuttall.flac
17 964 054 nuttall+gauss(0,1875).flac
17 964 127 blackman+blackman_harris_4term_92db.flac
17 964 130 blackman_harris_4term_92db+blackman.flac
17 964 996 blackman_harris_4term_92db+flattop.flac
17 965 003 flattop+blackman_harris_4term_92db.flac
17 965 025 gauss(0,125)+gauss(0,25).flac
17 965 030 gauss(0,25)+gauss(0,125).flac
17 965 071 blackman_harris_4term_92db+kaiser_bessel.flac
17 965 080 kaiser_bessel+blackman_harris_4term_92db.flac
17 965 141 gauss(0,0625)+rectangle.flac
17 965 141 rectangle+gauss(0,0625).flac
17 965 206 blackman+nuttall.flac
17 965 220 nuttall+blackman.flac
17 965 378 bartlett+gauss(0,0625).flac
17 965 380 gauss(0,0625)+bartlett.flac
17 965 474 gauss(0,5)+bartlett_hann.flac
17 965 478 bartlett_hann+gauss(0,5).flac
17 965 530 flattop+nuttall.flac
17 965 535 nuttall+flattop.flac
17 965 804 blackman+kaiser_bessel.flac
17 965 806 kaiser_bessel+blackman.flac
17 965 934 tukey(0,25)+rectangle.flac
17 965 940 rectangle+tukey(0,25).flac
17 966 054 kaiser_bessel+nuttall.flac
17 966 058 nuttall+kaiser_bessel.flac
17 966 442 tukey(0,5).flac
17 966 442 tukey(0,5)+tukey(0,5).flac
17 966 552 bartlett_hann+gauss(0,0625).flac
17 966 552 blackman_harris_4term_92db+gauss(0,25).flac
17 966 552 gauss(0,0625)+bartlett_hann.flac
17 966 557 gauss(0,25)+blackman_harris_4term_92db.flac
17 966 840 tukey(0,375).flac
17 966 840 tukey(0,375)+tukey(0,375).flac
17 966 915 tukey(0,625).flac
17 966 915 tukey(0,625)+tukey(0,625).flac
17 967 150 tukey(0,75).flac
17 967 150 tukey(0,75)+tukey(0,75).flac
17 967 306 nuttall+gauss(0,25).flac
17 967 307 gauss(0,25)+nuttall.flac
17 967 789 gauss(0,1875)+gauss(0,25).flac
17 967 789 gauss(0,25)+gauss(0,1875).flac
17 968 248 nuttall+blackman_harris_4term_92db.flac
17 968 257 blackman_harris_4term_92db+nuttall.flac
17 968 380 flattop+gauss(0,0625).flac
17 968 380 gauss(0,0625)+flattop.flac
17 968 521 flattop+gauss(0,25).flac
17 968 522 gauss(0,25)+flattop.flac
17 968 665 welch.flac
17 968 665 welch+welch.flac
17 968 686 rectangle+tukey(0,125).flac
17 968 688 tukey(0,125)+rectangle.flac
17 968 815 tukey(0,25).flac
17 968 815 tukey(0,25)+tukey(0,25).flac
17 968 906 tukey(0,875).flac
17 968 906 tukey(0,875)+tukey(0,875).flac
17 969 067 gauss(0,0625)+gauss(0,1875).flac
17 969 072 gauss(0,1875)+gauss(0,0625).flac
17 969 384 gauss(0,4375).flac
17 969 384 gauss(0,4375)+gauss(0,4375).flac
17 969 405 connes.flac
17 969 405 connes+connes.flac
17 969 766 hamming.flac
17 969 766 hamming+hamming.flac
17 970 019 triangle.flac
17 970 019 triangle+triangle.flac
17 970 089 hann.flac
17 970 089 hann+hann.flac
17 970 259 tukey(0,125).flac
17 970 259 tukey(0,125)+tukey(0,125).flac
17 970 439 flattop+gauss(0,125).flac
17 970 440 gauss(0,125)+flattop.flac
17 971 012 gauss(0,375).flac
17 971 012 gauss(0,375)+gauss(0,375).flac
17 971 453 gauss(0,125)+gauss(0,1875).flac
17 971 453 gauss(0,1875)+gauss(0,125).flac
17 971 846 gauss(0,5).flac
17 971 846 gauss(0,5)+gauss(0,5).flac
17 972 298 blackman.flac
17 972 298 blackman+blackman.flac
17 972 560 flattop+gauss(0,1875).flac
17 972 564 gauss(0,1875)+flattop.flac
17 972 772 kaiser_bessel.flac
17 972 772 kaiser_bessel+kaiser_bessel.flac
17 974 126 blackman_harris_4term_92db.flac
17 974 126 blackman_harris_4term_92db+blackman_harris_4term_92db.flac
17 975 537 nuttall.flac
17 975 537 nuttall+nuttall.flac
17 976 165 gauss(0,25).flac
17 976 165 gauss(0,25)+gauss(0,25).flac
17 978 186 gauss(0,0625)+gauss(0,125).flac
17 978 189 gauss(0,125)+gauss(0,0625).flac
17 979 510 gauss(0,1875).flac
17 979 510 gauss(0,1875)+gauss(0,1875).flac
17 980 074 flattop.flac
17 980 074 flattop+flattop.flac
17 986 239 gauss(0,125).flac
17 986 239 gauss(0,125)+gauss(0,125).flac
17 999 511 gauss(0,0625).flac
17 999 511 gauss(0,0625)+gauss(0,0625).flac
18 022 699 rectangle+bartlett_hann.flac
18 022 700 bartlett_hann+rectangle.flac
18 027 729 rectangle+bartlett.flac
18 027 730 bartlett+rectangle.flac
18 035 466 bartlett_hann+bartlett.flac
18 035 468 bartlett+bartlett_hann.flac
18 048 318 bartlett_hann.flac
18 048 318 bartlett_hann+bartlett_hann.flac
18 060 439 rectangle.flac
18 060 439 rectangle+rectangle.flac
18 072 654 bartlett.flac
18 072 654 bartlett+bartlett.flac

Synthetic Soul

May 4 2006, 09:04

I've got -3 to -6 into the spreadsheet so far, and apart from -3 (0.9) the other runs are pointing toward tukey(0.7) being the best value for my test corpus.

One interesting point is that if I calculate the average compression rate for the files between 50% and 70% compression at -5 (cutting 8 files lower and 9 higher) the best average compression rate is actually achieved using tukey(0.9), not tukey(0.5) as I might expect. This shows that both high and low compressed files prefer a lower value, dragging the value down from 0.9 to 0.7 in my corpus.

The corpus is too weighted toward pop/rock. This is evident by the mean compression at -5 being 60.360% and the median being 65.181%. I did purposefully choose files that were diverse, but weighted toward my general preference (rock). In my opinion many users' music collection falls in this area. However, in retrospect, the results may have more useful if the corpus was properly distributed between high (quiet/classical) and low (heavy rock) compression songs. This would give a more "linear" return, and ensure that averages were just that.

So, what do you do?

It depends on what the majority of users listen to; for me it would be tukey(0.9), for a classical music fan it may be tukey(0.2).

I'd be tempted to go with a higher (than 0.5) value (e.g.: 0.7) on the assumption that many users' music falls in the 50-70% (pop/rock), but that's one for Josh to call. In fact, I've just calculated the optimum for the files in the 45%-75% range and 0.7 is still top. Josh will know best what sort of percentage of files fall within this 45-75% range. I would suspect that it is very high though.

FYI, for those who won't bother to look at the spreadsheet (and who can blame you), here are the tracks that fall in the 50%-70% group:

CODE

Wagner                Ride of The Valkyries
Michael Kamen         Band of Brothers Suite Two
Corinne Bailey Rae    Choux Pastry Heart
Gorillaz              O Green World
Corinne Bailey Rae    Put Your Records On
Youssou N'dour        Alu Jon Jonki Jon
Paul Weller           The Changingman
AC/DC                 Shake Your Foundations
Skin                  Purple
The Kooks             Naive
The Magic Numbers     Forever Lost

Also note, averaging the files between 55% and 75% (10% either side of my median) also suggests tukey(0.9). This group starts to include such gems as Motorhead's "Ace of Spades", and Block Party's "Like Eating Glass", which are a lot more on the rock side of things (just the way I like 'em). So, in essence, it appears that only the far ends of the scale of my corpus is bringing the value down to 0.7.

Synthetic Soul

May 4 2006, 10:03

QUOTE(Synthetic Soul @ May 4 2006, 03:04 PM)

I'd be tempted to go with a higher (than 0.5) value (e.g.: 0.7) on the assumption that many users' music falls in the 50-70% (pop/rock), but that's one for Josh to call. In fact, I've just calculated the optimum for the files in the 45%-75% range and 0.7 is still top.

This obviously relies on the fact that tukey is actually the best algorithm to go for with files in this range. It may be that when you get into areas where tukey(0.7) provides the best tukey compression other algorithms may do better anyway. I.e.: tukey may do best in comparison at 0.6 or lower...

I'll test tukey(0.5) and tukey (0.7) against welch; hamming; connes; hann; blackman; and triangle next.

I've been looking at everyone's reports and I'm finding it so difficult to make any conclusions.

jcoalson

May 4 2006, 11:08

QUOTE(Synthetic Soul @ May 4 2006, 11:03 AM)

I've been looking at everyone's reports and I'm finding it so difficult to make any conclusions.

yes, but one good thing is that the results are very close (just a small fraction of a % between top contenders). also I can see a pattern where certain windows are better for higher filter orders. the mono result is also interesting.

I think when synthetic soul's single window test wraps up that will be enough data for single windows. the only other thing I'd like to see is a few more results for 2 window combos for -5 and -8, and maybe a few more mono tests.

thanks everyone for the tests so far.

Josh

Synthetic Soul

May 4 2006, 11:39

The tukey spreadsheet is now complete.

There is little to speculate different to my ramblings above.

The comparison with other algorithms has just started. I may be able to post that later tonight, or it may have to wait until tomorrow morning.

Edit:-3 to -5 completed and so far hamming is a clear winner. Unfortunately -6 to -8 will take a lot longer to come through.

flac-overall-results.xls

Edit 2: To be more precise, hamming is a definite favourite for 0-70% compression, while tukey is better at 70%+.

Cartman_Sr

May 4 2006, 13:13

Give me something to test. Seriously, I'm leaving the computer soon and I can set it up to do some tests if you guys want. Should I do some of the multi-function window tests (maybe with Egor's script)? Or if you have those good batch files you're using, Synthetic Soul, I could do those ones. I have some cpu time to spare.

Synthetic Soul

May 4 2006, 13:52

If you're still there, I'd go with Egor's.

I don't yet have scripts for multiple A values, and I'm a little busy at the moment to write one.

Good luck!

Edit: Update on flac-overall-results.xls. hamming was still a clear winner up to -6, but I've just got the results for -7 and tukey takes it; hamming did very poorly.

I actually wonder whether hann should be used for low compression (70%+), because tukey, according to my other spreadsheet, changes at around 80%+ to require lower values. So, rather than hamming; tukey(0.7); tukey(0.2), it may be easier to just use hamming; hann, if Josh intends to change the algorithm according to "spectral characteristics". If not, then by my reckoning its hamming for -3 to -6 and tukey for -7 to -8...

Take a look at the figures and make your own assumptions...

Synthetic Soul

May 4 2006, 15:21

OK, both tests have been completed.

flac-tukey-results.xls

flac-overall-results.xls

I'm not sure about the hann suggestion, maybe I would just say: -3 to -6: hamming; -7 to -8: tukey(0.7).

Again though, I'd be interested to see others' interpretation of the results (mine, and everyone else's).

Cartman_Sr

May 4 2006, 18:43

Geez this is taking forever. I'll post the results as soon as its done...

Cartman_Sr

May 4 2006, 21:17

Click to view attachment

Finally. I actually had to stop the script because I think it may have been caught in an infinite loop or something. Anyway, here are the results I got. The test file was track 3 from Star Wars Episode 1 soundtrack, and is classical music (instrumental, not too "complicated"). The results show that there is a definite improvement over the standard -8 level of encoding.

CODE