It's been a busy year in lossless audio codec land

So I thought it might be the right time to throw in another favourite codec poll, now that the lossy counterpart is into v2 and the last multi lossless codec poll is just a few days short of one year old. The latter included neither Apple Lossless, which has rapidly gained quite some popularity due to its support in iTunes, nor TTA.

My superficial impression tells me that OptimFROG, TTA, WavPack and WMA Lossless have also been gaining momentum ever since, perhaps at the expense of Bonk, LPAC, RK Audio, and others, which seem to have become obsolete and have therefore been omitted from the poll choices (although the board's 10 answers max has something to do with that as well). Feel free to mention LPAC or whatever you might still be using in a separate post below. Needless to say any other comments are most welcome.

Curious to know what the current balance of power between FLAC and Monkey's might be. Would Shorten still be able to keep a grip on a loyal share of users? How widely are OptimFROG and WavPack's hybrid modes being used? And what about rather obscure RealAudio Lossless?

In short: looking forward to your replies

Edit: grammar.

BTW, I'm using primarily FLAC for playback and occasional transcoding to lossy, because of its wide support and compatibility, cross-platformness, and fast decoding. Its open source character is a plus too. What I don't care about is fast encoding, since that's only done once, so I encode my FLACs @ -8.

I have previously been using Monkey's Audio, but a couple of personal negative experiences with file corruption, along with reports acknowledging this, have led me to lose faith in Monkey's Audio as a format.

I have recently taken up the rather big task of making an extra La (-high -noseek for maximum compression) back-up of all of my FLACs on an external hard disk, which I also apply PAR2 to. The Las are just for archiving - i.e. not for playback or any other use, just for disaster recovery - and for occasional internet transfer (if the receiver is on MS Windows).

Edit: italics.

I voted the same as last year: FLAC
mainly because of the wide software support, the good compression ratios and the wide range of systems it is supported on

I prefer Apple Lossless!
Because its supported in iTunes on my Powerbook, as well as my iPod mini and AirPort Express.

Monkey's Audio "normal": more efficient than Flac, and with correct decoding speed; conveniant tagging system for my usage (I can add as much information as I want, without juggling with padding values); MD5 for very fast verification (useful on disk transfer).

Monkey's Audio Extra High

i vote for flac simply because it has greater usability support on linux then the other codecs.

i do use monkey's audio, shorten, optimfrog and recently wavpack for testing.


later

FLAC for me ... although Wavpack 4 is also quite nice.

Right now I'm waiting to see if the Squeezebox gets some sort of lossless support in hardware. If so, I'll invariably go with whatever they choose.

WavPack for me. I've used it for a while, and now that 4 is released it's even better.

Flac for its so easy to use, tag and transcode. I have never experienced any corruptions or crc errors.

Although, I miss good compression ratio on flacs using 24 bit 96 khz. I get far better compression in theese rates uing winzip or winrar. Anyone know more about this?

Everything I rip gets backed up to Monkeys Audio - Extra High. My main reason's for using Monkeys is the compression ratio's (and the irresistible Monkey).

FLAC and LPAC.....and when I have bought a 250gb+ harddrive, I will use Monkeys Audio Extra High just as evereux

Same as evereaux for ripping, but I'm pretty sure Extra High is a virus eating away at what makes Monkey's Audio any good. The monkey really is irresistable, though. 3.97 High for me, until dBpowerAMP gets support and people stop being angry with me for sending files encoded with 3.99.

I'd just like to add (although I should maybe have done this in my opening post in this thread) that, should you choose to post a reply, it'd be nice if you could be as elaborate and specific as possible as to why you prefer whichever codec.

A mere I like this or that because it's the coolest! can hardly be called informative. If there's nothing else you'd specifically like to say beside indicating your preference, well, you can express that sufficiently just by picking one of the options in the poll itself. Apart from an overview of the current balance of power between codecs, to get to know people's specific motivation is what makes this poll and its comments interesting and instructive.

To sum it all up: thanks for taking the time and effort to write more than the proverbial word or two.

OptimFROG of course. It can be most efficient in terms of compression or transparently fast, it has a dual stream feature, and is under serious development (further nice features incoming). The only bad thing that comes to my mind is that AFAIK it is available only for Windows and Linux at the moment, and isn't supported on portables, but personally I don't care about that at all.

Monkey's Audio is going cross platform though: there are already plugins for XMMS and GStreamer (the future multimedia platform for GNOME). However, I haven't seen any initiative for MacOSX.

I use Monkey's Audio and FLAC for specific albums. Sadly, for some unknown reason ASIO and APE don't agree in terms of gaplessness so when going from one track to another there's an almost stuttering effect. If I switch to wave out or ds the gaplessness remains with no stuttering, however, if I use FLAC and ASIO it's completely gapless so that's what I use FLAC for, albums in which gaplessness is imperative. FLAC is a lil faster on the decode time and seeking, but I simply adore the embedded cuesheet feature in Monkey's Audio. It's so clean and simple to have a single file album and if you want an individual track or two real fast out of the image all you need is foo_monkey and a couple seconds. If FLAC had that feature I would use it exlusively and if Monkey's Audio was gapless with ASIO I would use that exclusively. Oh, well you can't win them all I guess.

BTW, I voted for Monkey simply because I've encoded so many more albums to it, but I still use FLAC a lot.

I thought foo_flac does now; check with case.

Josh

Thanks for the fast reply. Well, I'm still on foobar 0.8.1 for some reason, mostly due to things being so hectic (and a lil procrastination thrown in for good measure of course) I haven't had a chance to switch so I'm still using foo_flac 1.0.7 and I'm not sure if there's a new one in the latest versions of foobar. Also, I'm not sure if you were referring to the embedded cuesheet feature or quickly encoding a track from within a cuesheet. foo_flac is a decoder whereas foo_monkey is an encoder (foo_ape being the decoder) and I apologize in advance as I'm sure you already knew this I'm just a bit confused as to which problem could potentially be solved with foo_flac and it's also very early for me so I will go wake up before I post anymore.

Thanks again for the reply and thank you very, very much for your work on FLAC, it's appreciated more than you know.

ah, I see. I was referring to the decoding/playing of indivdual tracks from a FLAC + cuesheet... sorry for the mixup.

Josh

FLAC for several reasons

Open source, which was always one reason I liked Vorbis
Fast decode and low CPU on playback
Extremely easy to mass encode FLAC to Vorbis with oggdropXPd and keep my tags for use on my iRiver.

Momentarily I am using Monkey's Audio standard.

Good file size and perfect internal cuesheet support with foobar2000 are what won me over.

Although I am currently thinking about switching to wavpack. It seems a very nice compromise between compression ratio and decoding speed for me. Also internal cue sheet support works just as well with foobar2000 thanks to Case. Furthermore I just somehow apperciate all the work Briant has put into this and was very impressed by the new 4.0 release.

Ok...why FLAC and LPAC....good decoding time, good encoding time, and good compression ratio. FLAC is good because of the open source too.

And while foo_flac is the counterpart of foo_ape the decoder, foo_flaccer is the counterpart of foo_monkey the encoder.

I use FLAC too, for its cross-platform support and Freeness (as in freedom.)

Wavpack because I also use wavpack as an hybrid compressor and because it is fast enough for me.

Apple Lossless since it integrates nicely with iTunes. I don't care about vender lock-in: It's lossless, so if I ever need to switch format, I can just re-burn the DVDs.

Someday, I'm going to do a comparison between Apple Lossless and FLAC. I suspect that Apple Lossless is a lot faster on macs.

WavPack. Because it offers good compression in 'high' mode at reasonable speed, uses APEv2 Tags (no more padding), has a lovely foobar2000 plugin with CUE sheet support and is cross-platform.

WMA. I feel a bit more secure than with Monkey Audio, although as technology Monkey Audio is the best IMO. Everything works equally well with foobar and I like to use WMP for tagging sometimes. WMA's guaranteed to be easy. I think WMA will be a popular codec once lossless ripping goes mainstream, particularly if WMP ripping improves. And I expect to keep my non-lossless portable player for the next few years (though the Rio Karma with FLAC is attractive).

FLAC.

FLAC because as a few of you have said its open source. I also like vorbis for that reason and am in the process of reripping all of my CD's into FLAC from vorbis.

i use flac for the most part but i will probably change to wavpack. it's encodes 15% quicker than flac on my PII 400 and decodes 10% faster. the file size change is minimal although in 90% of samples i tested wavpack files were smaller than flac by 200 to 1000 kb. i play all my lossless files on my computer so extensive hardware support is not so important.

I've been using Ape for 3+ years mainly because:
-EAC fully supports macdll.dll to rip and tag on the fly;
-Better compression, in normal/high mode, than FLAC, etc;
-Very fast compression/decompression, even in older machines;

Have you actually timed this? Which setting do you encode your WavPacks with?

FLAC to archive.

LA - maximum compression for lossless archiving.

(I use Musepack for playback on computer and MP3 for my portable player)

Perhaps it's time for some reflection.

What's your lossless codec of choice?



(brackets) = last year's poll.
Note to HA's admins: what's with IPB's disproportional % bars? Shorten's 0% looks almost = OptimFROG's 2.2% and WMA's 2.2% looks almost = 1/2 WavPack's 10.6%.

This poll may not be a statistically entirely foolproof survey, but it does indicate certain tendencies. So here are my personal conclusions.

• The big three, and the only three ending in double digit percentages, remain FLAC, Monkey's Audio and WavPack, in descending order.
• Obvious winners are WavPack and Apple Lossless, and to a lesser extent also OptimFROG and TTA.
• Monkey's Audio and Shorten, and to a lesser extent La, come out as losers.
• One of the reasons that might explain this poll's winning codecs' (increasing) popularity among members of this forum is their respective developers' visible, though never obtrusive presence here and their enthusiastic, benevolent user support, at the expense of those who rarely show their face here (when was the last time we read something relevant from Michael Bevin (La developer) and Matt Ashland (from Monkey's Audio)?). Consider this a strong plea for active developer-end user interaction. It would be a great pity if excellent compressors like La and Monkey's were to fall behind.
• The success of both FLAC and especially Apple Lossless illustrates the importance of hardware support. Neither of them offers particularly strong compression compared to La, Monkey's Audio and OptimFROG, or speedy encoding compared to Monkey's and OF's faster encoding settings. As a consequence however, FLAC and ALAC's brisk decoding is probably one of the major reasons for their employability on hardware devices.
• With FLAC retaining the top rank and even reinforcing its lead in popularity and use over the other codecs, getting over half of the votes in both polls and some 2/3 of all posts in this particular forum, perhaps the time has come to consider creating a separate FLAC subforum, along with or next to this one (cf some time ago the split-up of the Lossy Codecs forum into the AAC, MP3, MPC and Ogg Vorbis subforums that we know now).

Just my 2 cents

Oh, before I forget: thanks to all who've voted.

Edit: grammar.
Edit 2:
- Updated my results image, old one is still available here;
- The Lossy Codecs forum split-up precedent statement turned out to be unverified, or misunderstood on my behalf to say the least, sorry about that!
Edit 3: Another update of the results graph; the 227 votes version is still here and the 264 one here. BTW, the 2003 results in my graph are frozen to the 164 votes state in which they were at the beginning of the August 2004 poll.
Edit 4: Update of the results image. Previous one (/312) is here. Seems like WavPack's dropped to a single digit now...

I don't think that particularly reflects in popularity. Monkey's Audio has always been a popular encoder, and Matt posted here precisely once. Ghido, on the other hand, has posted several times, and still his codec is used as much as WMA Lossless, whose developers never showed up around here.



Sorry, I don't buy it. That (hardware support) might be one of the reasons, but there are others to consider. FLAC gained great momentum because for a long time it was the only truly multiplatform lossless codec (other than shorten, but... heh), and also because it decodes very fast, which is an added bonus for people that plan to transcode to MP3 for their hardware players. It was also the only OSI-approved codec, which caters to the countless open source zealots among us.

And I don't agree Apple Lossless is successful. Less than 5% doesn't sound like that to me, at least. I think a good amount of these 11 members that prefer ALAC are using it because they are on Macintosh (which has no decent lossless alternative) and not because of hardware support.



There was never a "Lossy Codecs" forum; AAC, MP3, Vorbis and MPC have always been separated.

And I don't think it's justifiable to split the lossless forum in "Flac" and "Everything else". There simply isn't enough discussion going on about lossless to justify a split, I think. Lossless discussion is naturally limited because there's no point discussing subjects like quality and listening tests, like happens in the lossy forums. And also, because most people are still on lossy.

Regards;

Roberto.

that would help me keep up on the FLAC topics for sure! but it's cool either way, I do the occasional shotgun search to try and keep up.

the results are interesting. kind of like watching one of those carnival games where everyone's squirting water into a target to make their horse/car go faster.

it would be nice in the next poll to have 'other' and 'don't use lossless' split so we could normalize the percentages against actual lossless users, or better yet have no 'don't use' entry all since that could be covered better by a different poll.

Josh

Maby I'm a bit late on giving a comment about the lossless codec I use but here goes.
I use Flac at cause nothing decodes fatser, except shorten, than Flac. And using any other lossless codec that has a better compression will only make file a few percent smaller and a lot slowwer to decode. I use level 5 cause it gives me a decent compression and fast compression. I dont understand why some use level 8, files encode a lot slower and files dont get any smaller. At level 0 it performs quite simmilar to shorten. I'd love to se flac keep on developing.

Yes, -8 does encode a lot slower than -5 (= default), 4 to 5 times as slow to be more specific, but that doesn't have the slightest effect on decoding speed. That's the beauty of FLAC, no matter what encoding level you pick, decoding won't be influenced.

So that's why I encode at -8. As long as it squeezes out every byte it can (compression ratio difference between -5 and -8 may not be much, but every kB counts) and doesn't affect decoding time, I don't care if encoding takes a proverbial year. Since you only encode once, and encoding is not something I sit and wait for, but is just a background task, how long it takes matters zip to me.

flac --super-secret-totally-impractical-compression-level

Alright, I asked for it
Should've known someone was gonna throw that at me
Seriously though, FLAC's --super-secret-totally-impractical-compression-level does affect decoding speed, contrary to all of the other encoding levels.

Still, it is my impression that this poll's winning (advancing) codecs' developers are all HA regulars. It can't be claimed that FLAC, OptimFROG, TTA and WavPack are suffering from Josh, Ghido, Alexander and David's presence here, like La and Monkey's can't be said to be benefiting from their respective developers' absence from this board.

Oh absolutely. Just because I didn't express myself in conditional clauses, that still doesn't mean that I stated otherwise (that there are other reasons to be taken into account, I mean).

It does to me. After all, Apple Lossless wasn't even around some 4 months ago. Taking the No 4 spot with 5%, coming from scratch in less than 4 months' time looks like quite a performance to me.

If I didn't know any better, I'd think you were trying to make it sound like iPod and AirPort Express support is actually a disadvantage.

Right. That's proved to be quite an imprudent claim on my behalf.

I'll address that in the Site Related Discussion thread.

Exactly. They aren't suffering nor benefitting. As I see it, developers participating in HA make no difference on their codecs' popularity.

If there was an influence in popularity, people would be migrating from Vorbis and MPC (their developers barely post here anymore) to AAC and Lame.



I don't agree, precisely because it had a head-start - the Macintosh user base, that as the situation is now won't use anything else. If it had no previously sympathetic user base, I doubt it would have 5% (5 in each 100, which, for me, is veeery few)

One could extrapolate what you said and claim "with this adoption speed, ALAC will beat FLAC in a few months". Of course, that makes no sense. The codec had a head start, and now it's to be expected that it'll more or less stabilize at a certain market share.



Nah. I'm just saying that you are giving hardware support more credit than it deserves.

I'd have to disagree with you here. The decoding speed does not get affected much, but it is still affected. I did a big ol' test on my computer, and got the following results (the UCSC server seems to be passing a brainstone... be patient):

Zoomed horizontally:

Horizontal axis is file size ratio (lower is better) and vertical is decoding speed in "X" (higher is better) I did this test on 8 different songs, decoded 12 times, then threw out the highest and lowest decoding times. The points, then, are the average of 80 trials. The "-5" label looks wierd because I also tested flac with no options, and called it "--". Well, they're indeed the same, so the label is both "-5" and "--". "SS" is --super-secret-yada-yada.

The test was done on a P4 1.5ghz with 512MB RAM, using the foobar diskwriter speed test.

BTW, -2 gave consistently higher decoding speeds than either -1 or -0. Not sure why...

Also, note that the horizontal scale is greatly magnified. The difference between -0 and -8 is only 5% the original file size.

Based on the results of this test, I'm using -7. -7 decodes at esentially the same speed as 4-6, but has higher compression. If anybody has different results, I'd love to hear them.

Hmm if it's placed at this spot it's not because of its technical merit. It's only because Apple limits the user's choice and there's yet no decent player in Apple platform supporting other lossless codecs.

I don't imagine it could score much higher. People don't care much about lossless, to them 128kbps mp3 is cd quality anyway. And the rest of the users who would be interested in lossless codecs know what they want: they wouldn't want to use an inferior codec unless their choices are limited.

FLAC had been used on the Mac for some time. It was THE lossless choice on the OS X side with MacFlac and Flacer encoding and VLC and (no longer developed, but great) MacAmpLiteX for playback. Then ALAC came out. I certainly switched.

I'd say hardware support is the much bigger reason for its success. With the still very well selling iPod, to some peoples mind, it's not that ALAC is the only viable lossless on the Mac, its that ALAC is the only viable lossless period... for home and away.* I expect it to get more popular quickly (it's after all for many their first experience with lossless).


* Yes, of course this is missing the Karma and FLAC.

In my opinion, they do. After all, this is a Hydrogenaudio poll, and every single voter is an HA member. It can therefore be reasonably assumed that they read HA and, more specifically, the lossless forum on a somewhat regular basis, where certain codecs get more exposure time than others, just from (the absence of) their developers' posts.
If I would have been asked for my favourite lossless compressor a year ago, when I hadn't even heard of HA, of course I would've answered Monkey's Audio. I simply didn't know any better, until I started reading HA about FLAC, OptimFROG and WavPack, which have been rising in my hit list ever since

Edit: to sum it all up, I'm not trying to say that developers posting in HA is of major influence on their codecs' popularity, but it can't just be dismissed.

I object. I've never hinted at that anywhere. Besides, I'm convinced that any prediction, in matters where so many factors are into play, is entirely senseless.

Intriguing work, Omion. Especially since yours is the very first ratio/speed test of --super-secret-etc-level I've come across. Thanks a bunch Are you planning on putting it online somewhere (apart from this thread, that is)? I for one would appreciate that.

Regarding those different results you'd love to hear about, well, they're the test results I've been relying on so far: Hans van der Heijden's test, the one by Wim Speekenbrink and the one on the official FLAC site (even though the latter might stricto sensu not be an independent one). All of them, as you'll be able to read, are quite comprehensive and report nominal difference in decoding speed between the various encoding levels.

Edit:
On second thought, the fact that, especially over a 12 times' decoding average, your -8 and -2 encodings gave such deviant results (well, one's gotta argue about something, right? ), might be attributed to the limited 8 song base. Hans van der Heijden's FLACed some 80 songs for his test, and each of the 8 compression levels he tested (including -8 and -2, but no --super-secret and -0) decoded at an average 51x real-time speed on his 900 MHz Athlon:

source: <http://web.inter.nl.net/users/hvdh/lossless/lossless.htm>

(Edit 2: Grammar.)