Hi, out of interest and I know it depends from the encoder: Just had a look into the compression values of my lossless files (all encoded with TAK from normal to max). I mainly have IDM (1/3) music files and Classical orchestral (2/3), 577 files to be precise. All in all, for classical music, my experience is that lossless compression is fun, it often clocks in between 400-600 kbps.
I was surprised electronic music does often compress worse, surprising for me.
But in case of my "worst compression" winner it's not surprising that it's an electronica track, because maybe of it's complexity.
My winner is: Autechre - Gantz Graf (971 kbps with TAK 5 max. /Tak v.1.03)
The loser isn't too interesting, because it's stuff with lots of silence (ae/hafler/"aeo3") followed by many classical tracks starting at 366 kbps...
What's your winner?
Hi, out of interest and I know it depends from the encoder: Just had a look into the compression values of my lossless files (all encoded with TAK from normal to max). I mainly have IDM (1/3) music files and Classical orchestral (2/3), 577 files to be precise. All in all, for classical music, my experience is that lossless compression is fun, it often clocks in between 400-600 kbps.
I was surprised electronic music does often compress worse, surprising for me.
But in case of my "worst compression" winner it's not surprising that it's an electronica track, because maybe of it's complexity.
My winner is: Autechre - Gantz Graf (971 kbps with TAK 5 max. /Tak v.1.03)
The loser isn't too interesting, because it's stuff with lots of silence (ae/hafler/"aeo3") followed by many classical tracks starting at 366 kbps...
What's your winner?
Reef - [2000] Getaway\02 - Reef - Superhero.flac = 1133.1kbps (FLAC -8).
(same file, lossyWAV -3 processed, FLAC -5 -b 512 = 374.6kbps )
Would you mind testing this file with other codecs, such as flac, WavPack and Monkey's Audio for comparison purposes?
I'm not personally interested in ALAC, WMAL, TTA or OptimFrog, but others might be.
I assume only CD music counts.
Modern Talking - [2001] America - 04. SMS to My Heart
FLAC 1133 kBit/s [80%]
Very loud (compressed) recording.
No surprises.
Gantz Graf:
- with "reference libFLAC 1.2.1 20070917" at -8: 1065 kbps
- with latest Wavpack at extra high (-m -i -q -hh -x6): 1025 kbps
I have no monkeys encoder/never used it.
I reckon compressed "loud" music with as much as possible use of signals over the whole frequency range is what counts?
And the award goes to... the white noise, of course!
Edit: could anyone try this out please?
I have no monkeys encoder/never used it.
Pity. I'm betting the results will be lower than TAK using the Extra High setting, if not the High setting.
Care to give it a shot (http://www.rarewares.org/files/ogg/MAC.zip)?
mac.exe <Input File> <Output File> -c3000 (for High)
mac.exe <Input File> <Output File> -c4000 (for Extra High)
Pity. I'm betting the results will be lower than TAK using the Extra High setting, if not the High setting.
Care to give it a shot (http://www.rarewares.org/files/ogg/MAC.zip)?
OK, Gantz Graf with monkeys at -c4000: 963 kbps, very quick encode. But searching within tracks is pretty slow with monkeys?
And the award goes to... the white noise, of course!
Edit: could anyone try this out please?
Interesting. 1 Minute of white noise created with Magix Samplitude: 679k average with TAK@pmax. EDIT: But the generated noise is at 50% volume. The normalized, 100% volume level white noise is at 812 kbps tak.
Atari Teenage Riot - Revolution Action (Various Artists - POP 2000)
FLAC 1.2.0 -8: 1167 kbps
FLAC 1.1.4 -8: 1165 kbps (interesting, isn't it?)
MAC -high: 1150 kbps
OK, Gantz Graf with monkeys at -c4000: 963 kbps, very quick encode. But searching within tracks is pretty slow with monkeys?
Perhaps a very slight delay with Extra High, but not with the High setting (do you have a number for this?).
You'll probably think the decoding is lethargic, and it is compared to the other three codecs.
Thanks for trying this out.
Venetian Snares + Speedranch - Halfway Up the Stairway of Mucus
1295 kbps with FLAC 1.2.0 -8
RG track gain = -16.26
I suppose only someone like Merzbow could come close to this.
60s white noise created by Audition (strength 40, uncorrelated channels) had 1386 kbps.
OK, Gantz Graf with monkeys at -c4000: 963 kbps, very quick encode. But searching within tracks is pretty slow with monkeys?
Perhaps a very slight delay with Extra High, but not with the High setting (do you have a number for this?).
Gantz Graf/Monkeys c3000: 982 kbps. Encoding maybe 8 seconds . Seeking inside this file is good and I must admit at my previous test I had gapless crossfader in my dsp chain. Good stuff. Does it support piping?
If the noise can be compressed, it's not white enough.
Uh, heavy metal routinely makes it well into the 1000+ kbps...
Last song on my playlist, for instance... Dimmu Borgir, The Invaluable Darkness: 1158 kbps, TAK 1.0.2 High Max....
Gantz Graf/Monkeys c3000: 982 kbps. Encoding maybe 8 seconds . Seeking inside this file is good and I must admit at my previous test I had gapless crossfader in my dsp chain. Good stuff. Does it support piping?
The version I linked supports decoding to stdout and encoding from stdin, but these do not work with foobar2000 for some reason.
http://www.hydrogenaudio.org/forums/index....st&p=540026 (http://www.hydrogenaudio.org/forums/index.php?s=&showtopic=60308&view=findpost&p=540026)
Thanks for posting the results for MAC High.
If the noise can be compressed, it's not white enough.
True zen.
Besides some modern clippressed albums made by loudness war, which go over 1000 kbit/s with flac -8,
I recommend those DTS CDs.
a guns n roses best of in 5.1 dts reaches 1264 k,
the DTS Santana - Abraxas gets 1256 k !
But still a noticeable compression compared to 1411 k.
Of course, the DTS CDs consist of nearly white or pink? noise in wav 16 bit, 44.1 kHZ stereo PCM format.
If the noise can be compressed, it's not white enough.
True zen. :-)
That made my day. Thank you! :-)
Besides some modern clippressed albums made by loudness war, which go over 1000 kbit/s with flac -8...
I listen mostly to rock and pop (almost all kinds) and Apple Lossless can compress sometimes above 1000 kbps. I noticed that the louder the music is the higher the compression rate goes and larger the file size is.
Mono files however compress around 600 kbps (probably because of some kind of joint stereo technique being used.)
If the noise can be compressed, it's not white enough.
http://www.hydrogenaudio.org/forums/index....showtopic=52490 (http://www.hydrogenaudio.org/forums/index.php?showtopic=52490)
Venetian Snares + Speedranch - Halfway Up the Stairway of Mucus
1295 kbps with FLAC 1.2.0 -8
RG track gain = -16.26
I suppose only someone like Merzbow could come close to this.
60s white noise created by Audition (strength 40, uncorrelated channels) had 1386 kbps.
OMG!!! Well that definately shows the correlation between noise and heavy compression
Can't beat yours, but my winner is (House): Studio B: C'mon Get It On, 1166 kbps with FLAC 1.2.0 -8, RG: -10.41. Yet, it's quite nice compressed, but it's sound allows for heavy compression.
Edit: Why not include best compression?
Enya: From Where I Am, 356 kbps, though peak is very low: 0.23, +10,53
Röyksopp: Tristesse Globale, 380 kbps, with peak and RG 0.73, -1,13 seems a fair comparison
Edit: Why not include best compression?
There are a lot of albums out there with minutes of silence in the last track, separating the "officially" last song from a "hidden track". The longer the silence, the better the compression.
For example, i have "Nirvana - All Apologies" off "In Utero" at 321 kbps (FLAC 1.2.0 -8) - with a length of 31:32 minutes.
However, I found only only album that has a separate track juist for the silence. So, beat this:
Track 13 off "Spillsbury - 2" - 13:13 minutes of pure silence
1 (one) kbps with FLAC 1.1.2 - I guess there is not use in trying other compressors :-)
Edit: Why not include best compression?
There are a lot of albums out there with minutes of silence in the last track, separating the "officially" last song from a "hidden track". The longer the silence, the better the compression.
For example, i have "Nirvana - All Apologies" off "In Utero" at 321 kbps (FLAC 1.2.0 -8) - with a length of 31:32 minutes.
However, I found only only album that has a separate track juist for the silence. So, beat this:
Track 13 off "Spillsbury - 2" - 13:13 minutes of pure silence
1 (one) kbps with FLAC 1.1.2 - I guess there is not use in trying other compressors :-)
Yeah, such like the white noise example you can always mess with the content to fit your claim. Additionally, you can cut off the silence start/end to get a higher rate - I didn't.
I DID check this before I posted my best compression rates - There's no silence.
My winner is: Autechre - Gantz Graf (971 kbps with TAK 5 max. /Tak v.1.03)
OT, but have you seen the video for that? Incredible. I wouldn't be surprised if it'd win worst video compression, too.
But yes, with classical, particularly with solo piano, I sometimes have to check if I'm actually listening to my lossless files based on the kbps.
Work Title : Black Album (Jpn 2006 Rem)
Performer : Metallica
Track/Movement Title : So What
Sample Rate : 44100 Hz
Channels : 2
Bits Per Sample : 16
Bitrate : 1106 kbps
Codec : Monkey's Audio
Codec Profile : Monkey's Audio Extra High
Now correct me if I'm wrong, but isn't it mathematicaly proven that average lossless compression rates cannot be better than 50%?
Isn't it also proven that bitrate can actually go up in worse-case lossless scenarios?
Isn't it also proven that bitrate can actually go up in worse-case lossless scenarios?
I haven't seen the proof for the first question, but I have for this one.
Now correct me if I'm wrong, but isn't it mathematicaly proven that average lossless compression rates cannot be better than 50%?
Isn't it also proven that bitrate can actually go up in worse-case lossless scenarios?
http://en.wikipedia.org/wiki/Lossless_data...ion#Limitations (http://en.wikipedia.org/wiki/Lossless_data_compression#Limitations)
Actually if we compress without loss *all* (there are 2^(n*8) different possible files of size n bytes) possible files the best possible average compression ratio is 0%, i.e. no compression. The best windows compressor in that case is the COPY command
However the files that the people use and compress (usually the one with meaningful content, unlike most files which are white noise) are a very small fraction of all possible files. Therefore a program can compress most actual files but it will increase the size of most possible files.
I was about to point out something obvious, but wikipedia had it covered already...
Any lossless compression algorithm that makes some files shorter must necessarily make some files longer, but it is not necessary that those files become very much longer. Most practical compression algorithms provide an "escape" facility that can turn off the normal coding for files that would become longer by being encoded. Then the only increase in size is a few bits to tell the decoder that the normal coding has been turned off for the entire input. For example, DEFLATE compressed files never need to grow by more than 5 bytes per 65,535 bytes of input.
Cheers,
David.
Best Compression:
Exhumed - Disfigured Corpse
161 kbps
TAK 1.0.3 @ TAK 5 Max
Worst Compression:
Nightrage - Release
1160 kbps
TAK 1.0.3 @ TAK 5 Max
Ha! Just a few week ago, I checked some of my songs out of couriosity which had the highest lossless bitrates, and whether it is possible to guess the bitrate by listening to it... Nice to see a thread for that.
Anyway:
Codec: Monkey Audio - High
Highest bitrate, 1st place: 1111kbps (Ayumi Hamasaki - Depend on You)
Highest bitrate, 2nd place: 1107kbps (Ayumi Hamasaki - Depend on You [different remix])
Lowest bitrate, 1st place: 290kbps (Arvo Pärt - Solitudine-Stato D'Animo)
Lowest bitrate, 2nd place: 302kbps (Arvo Pärt - Berliner Messe - Agnus Dei)
The "Für Alina"-Album by Arvo Pärt had even slightler lower bitrates, but I do not have those files anymore.
- Most strongly compressed files (up to 400kbps) are not necessarily silent, but have very little noise and sounds with relatively few overtones, and are slow in general.
- Uncompressible files (above 1000kbps) are typically highly compressed (sound weird, I know...) and fast, altought there are a few unexpected exceptions.
Just checked my small lossless collection, the maximum there is 1067kbps, Monkey's Audio (insane): Die Toten Hosen - Testbild (live). Replaygain is just -8.48. I don't use Insane anymore because it's too slow for the gain. The same file encoded with High is just 8kbps larger.
Brian Eno's Neroli compresses down to (coincidentally) 320kb/s with FLAC 1.2.0 -8, and there's no silence in the track.
Venetian Snares + Speedranch - Halfway Up the Stairway of Mucus
1295 kbps with FLAC 1.2.0 -8
RG track gain = -16.26
I suppose only someone like Merzbow could come close to this.
Ya think?
Merzbow: I Lead You Towards Glorious Times (from "Venereology"). Compressed with FLAC 1.1 -5 a while ago.
Bitrate: 1393kbps
RG track gain: -20.64db
I'd say Merzbow doesn't as much "come close to this" as it does "blow it out of the water".
60s white noise created by Audition (strength 40, uncorrelated channels) had 1386 kbps.
I knew it. Merzbow is less compressible than white noise.
thats because Merzbow is pure darkness
All files are FLAC --best and stereo except the last one, encoded using FLAC 1.1.4 and 1.2.1
Accuface – Re-Energize Your Soul (Tunnel Trance Force Vol. 14): 44100 Hz, 16 Bit, Track gain -6.05 dB, 1153 kbps
Jonah Dempcy – Mother Matrix Most Mysterious (http://www.archive.org/details/JonahDempcySyncreticBeliefs_4) (Synchretic Beliefs (http://revolutionvoid.com/rv007/)): 44100 Hz, 24 Bit, Track Gain -1.49 dB, 1215 kbps
Historic du Soldat: 96000 Hz, 24 Bit, Track gain -3.50 dB, 3000 kbps
Yes – Roundabout (6 seconds sample): 96000 Hz, 24 Bit, 5.1, Track gain -1.60 dB: 7766 kbps
To the idea: strong musical compression (limiting) --> less lossless compression:
Strong musical compression introduces lots of square waves which contain much energy in high spectral parts which is very similar to noise. That would explain the relation to compressibility.
To the idea: strong musical compression (limiting) --> less lossless compression:
Strong musical compression introduces lots of square waves which contain much energy in high spectral parts which is very similar to noise. That would explain the relation to compressibility.
Square waves are in fact quite easy to compress (just predict "this sample=next sample"). But not-quite-square waves that occur with limiting (without actual clipping) could be hard.
The low bitrates happen with files with many leading zeros on each sample (i.e. quiet!), but also with files with many trailing zeros on each sample (i.e. lossyWAV).
This way, you can make "normal" music get below 256kbps within FLAC, but it's no longer lossless - in fact it sounds quite horrible because though it says it's 44.1kHz 16bits stereo, there's only about 4 bits left! (This is a missuse of lossyWAV of course - default settings shoudl be transparent).
If anyone ever uses something like lossyWAV for a "sound effect" or a trick on a CD, you could have a normal (not quiet, not simple, not slow) yet very compressible track.
Cheers,
David.
It is far more likely that somebody could release a CD where the bit reduction is not the last step (it could be done before downsampling to 44100), or could pump the reduced data through an analogue link. This way the material would still sound horrible while being uncompressible.
Funny, best and worst awards belong to the same group. Wonder who?
Oasis - My Big Mouth [Be Here Now]: 1143 kb/s, 5:03, FLAC -6
Oasis - Better Man [Heathen Chemistry]: 319 kb/s, 38:03 (pure silence for the most part), FLAC -6
Of course, as you may expect, Oasis is all over the higher ranks of the list.
Both flac --best but different versions:
Terrorgruppe -- Videokamera: 1178 kb/s reference libFLAC 1.1.1 20041001
Pjotr Tschaikowski -- Piano Trio a-Moll op. 50, Tema con variazioni- Andante con moto (piano: Martha Argerich): 304 kb/s libFLAC 1.2.0 20070715
Funny, best and worst awards belong to the same group. Wonder who?
Oasis - My Big Mouth [Be Here Now]: 1143 kb/s, 5:03, FLAC -6
Oasis - Better Man [Heathen Chemistry]: 319 kb/s, 38:03 (pure silence for the most part), FLAC -6
Of course, as you may expect, Oasis is all over the higher ranks of the list.
That second example is kind of cheating.
That second example is kind of cheating.
OK, then let's try the next in the list:
U2 - 4th of July [The Unforgettable Fire]: 460 kb/s, 2:13, FLAC -6
Better?
Thus far I haven't seen anything of mine higher
Merzbow - Degradation of Tapes - 1208 kbps (WavPack; -h)
Generally speaking, most noise is going to get it done
I rip to images exclusively so the album with the highest lossless bitrate is...
The Darkness' Permission to Land @ 1063kbps w/ FLAC (reference libFLAC 1.2.0 20070715) -8
... I don't even like this CD.
There must be something higher on my network back home, I only have the recent CDs I've purchased on had as lossless.
001. Röyksopp - [The Understanding (Limited Edition) #05] Follow My Ruin
1040 kbps TAK 5 max
1066 kbps reference libFLAC 1.2.1 20070917
Worst Compression:
Give it Away, Red Hot Chili Peppers, Greatest hits album: 1206kbps. FLAC level 5
Best Compression:
Rondenas, Dennis Koster: 495kbps. FLAC Level 5
Wow, for Worst Compression I don't have anything in my collection anywhere near these others.
For Best Compression, do audiobooks count?
Some of the discs in the first Harry Potter book will drop below 300 kbps.
Worst with ALAC is Freestylers - Painkiller at 1102kbps. Best without any silence is Vespertilio from the Batman Begins OST at 545 kbps.
Best compression:
Bartók: Complete Piano Works CD 1 - Kocsis Zoltán (Phillips)
Flac -6: Album: 374 kbps (=26,5 %), Track: Track 6 (14 Bagatelles 6. Lento) 281 kbps (=19,91 %)
My best compression as of yesterday:
Beethoven's track 'Für Elise' using TAK -p4: 275 kbps.