I was wondering if anyone knows if there are plans for a flac 2, or are the developers happy with the current format?

I have no plans for a flac 2, which according to the numbering scheme would not be backwards compatible with flac. there would be no need for that unless flac was failing.

there will probably continue to be minor improvements but my focus for a while has been on making flac ubiquitous.

I am glad to hear that, as I would so much like to get a phone with Android being able to play flac files, and I would assume that would be easier to find (once Android becomes available), if the companies can see flac doesn't change.

That's at least the argument from the Vorbis developers.

It really amazes me how you can improve the compression and still be bit stream compatible!!

Keep the good work

Does it mean FLAC has nearly reach it's limit? (like mp3)
Since there's no FLAC 2. If you want MUCH better compression, you have to wait another future codecs.

AFAIK there won't be any codec with MUCH better compression. It's just not possible to compress audio much better without any loss.
Only lossy codecs (like AAC or MP3) really could be improved by optimizing the psymodel to the humen hearing.

Until now Mr. Coalson always found a way to make it better. Nothing of that needed a major name change like version 2.
I bet some minor improvements are again in the pipeline.

I would say breaking the bit stream to be able to gain another 2% or perhaps 5% better compression if no worth it.

If you have 100GB of flac music, and being able to compress it by yet 5% means you save 5GB. 5GB is ~10 albums. + you have to spend the time converting the 100GB. Is that 10 albums worth is?

I'd it is much more important the that hardware manufactures regonize flac as being bitstream compatible for a long time to come, so they dare to implement it.

The Vorbis developers made a promise from the very start that Vorbis 2 would be able to play on Vorbis 1 devices.

I don't see Lossless codecs getting MUCH better than they are right now... it's 99.8% perfect... I can't think of anything that will make FLAC and others compress another 35%+

depends on what you mean by "much". future flac encoders may get a couple more percent but I doubt 5% or more on cd audio. I doubt any practical codec will get more than 10% better than state-of-the-art right now on cd audio.

agree, this also explains why the few % advantage with some other codecs is not enough for most people to migrate from flac. in fact with flac's current momentum I don't think a flac2 could compete with it either. very few people will switch for a 5% or even 10% improvement. flac has 5% over shn but in the early days flac had to get much more useful before starting to displace it.

codec adoption is driven by the network effect and incompatible changes interfere with that.

i'm switching right now from TAK/MPC to FLAC/MP3... reason... there is no reason to swim against the tide...

Couldn't more compression come from a 'solid' compression mode, like that in the rar format?

I dread to think of how many moons it'd take to compress something with that though.

Even if there were repetitions in an audio stream, imagine how much memory would be required for encoding?

Try Flake and use the variable block size (method 1) switch. It's very fast and saves a couple megs on an album: flake -5 -v 1 input.wav -o output.flac
So far every FLAC I created this way was successfully tested with flac -t.

are you also switching from firefox to IE, and from whatever other DAP you have to an iPod?
honestly, i can't think of a worse reason to switch.

One problem would be that you would have to wait a couple of seconds before the songs starts, and that would only get even worse when you want to listen to an album from start to the end.

The point is that there is not enough of an advantage to TAK over FLAC considering how much more compatible FLAC is with media players and DAPs. Your analogy between firefox and IE is not a good one, since there are no compatibility issues running either one (under windows). The added functionality of firefox over IE means that I would choose firefox, since there is not a disadvantage to running it (only advantages). That analogy breaks down with FLAC and TAK precisely because of the compatibility issues; TAK is "better" than FLAC, but its not that much better, and TAK will play on foobar2000 and pretty much nothing else. Use Linux or Mac? No TAK for you! If you are only using foobar2000, then TAK is a good choice. But In my mind the transportability of FLAC is a huge advantage.

Mp3 vs other lossy codecs is another can of worms, since the differences between the formats are more striking. I use AAC because I think the quality difference is worth the compatibility problems. But mp3 is a perfectly acceptable choice because it just works.

AAC's advantage is that can achieve 'transparent' quality at a lower bitrate than MP3. It is even more noticeable in streaming radio where CBR is required. It is not, however, 'higher quality' as LAME MP3 is capable of achieving transparent quality in a vast majority of cases.

I think that OPERA was the pioneer in pretty much everything that IE and FIREFOX have incorporated. OPERA is my browser right now. It has speed dial, and it's much better and faster than FIREFOX (wanna check it? go back and forth when browsing the net, you will see). FIREFOX was slim in the beginning but it's just plainly bloated now. But leaving the browsers talk... I understand why a programmer won't release the source code, but I only see TAK's future is relying on pretty much what FLAC did to construct its user-base. So unless it is developed in Linux and Mac, unless people make plugins out of it, it can be dodgy. The hardware support is surely something to determine, and although in practice FLAC devices are a bit unaccessible (I would have to import ZIOVA or Squeeze-Box), it takes a little advantage on TAK.

well onto opera, i'd say that flac would be doing just as poor if it decided to be closed

Well I WAS switching to FLAC... got these "Error flushing file..." in FB2K.... LOL, and I can't do piping as well, because of the converter bug which will put 3000 seeks points or so in the file.

What I meant was that at the same bitrate, other lossy codecs sound better than mp3. But your right, "quality" is an ambiguous choice of words.

the problem is it's probably too late. flac filled the need at the right time the same way mp3 did. the only 2 things I think that can replace either of them are 1) agreement by all the big labels to distribute in some alternative format; 2) technology change in audio distribution (e.g. non-pcm coding like dsd)

note apple did 1) with aac but still it is not more popular than mp3

There has been some discussion on exploiting long-term correlation for lossy compression.

I guess the same holds true for FLAC: if one could assume infinite decoder memory and encoder processing power (both seems perfectly sane in the very distant future), then at least some kinds of music ought to be more compressible?

-k

How can you say that infinite decoder memory seems perfectly sane?
Our Sun will expand and destroy the Earth in the very distant future.

Mpeg4Als (lossless) has an option to exploit long-term correlation. It's very slow and the advantage is rarely bigger than 0.15 percent.

Well this is computer science, 2 is to a first approximation infinite

This is a very general claim that is not backed up by ABX results. If you compare codecs and wish to make specific reports about quality differences you have perceived, please back them up with ABX results to prove your point.

ABX does not prove quality. it proves differencesn.

there have been lots of tests on HA by Rjamorim. and to my memory it seems the AAC beats mp3 by a small margin in every bitrate where you could hear a difference.
but i might remember wrongly.

If there was infinite memory, there would be no need for compression. Not only processing power is evolving, but also is storage. Absolutely highest compression is rarely needed.

Let's take video as an example. Assuming infinite – for all practical purposes – processing power you encode your material in H.264. Then you purchase a silent low power computer. And guess what the videos don't play on it.

If we're discussing features that could be added to FLAC, I'd like to suggest the ability to work with 32-bit files like Wavpack currently does. If FLAC could support a 32-bit bit depth I'd not need any other lossless codec for archiving or listening.

Prior to FLAC, I was using Shorten which filled my needs at the time. As it stands I'm using FLAC where possible for compatibility and Wavpack when I'm working with 32-bit files.

Just my .02

sb

Yeah, FLAC 2 with floating-point support would be cool, as audio mastering is done in floating-point. That may be kind of extension over current codec for fp, and otherwise (for integer) fully backwards compatible. So to say, as long as FLAC 2 contains integer data, it conforms to FLAC 1 specification.

I'd come back to the Hydrogenaudio forums to find an answer to something google was horrible at doing. Which was are there any major online music stores selling albums in DRM free FLAC. The answer is of course no. There's a few notable artists and a few publishers, but no mass market retailers, as yet.

So the question I'd have for the future of FLAC is this.... Do you expect me to be able to buy DRM-free FLAC albums in the near to mid future?

I have a Cowon X5L, which means no DRM support. I've just got a Cybook Gen3 which means I've gone paperless for books. While I was tossing them all out, I put all my CDs together in a box, and they're now stored. I expect the only time they'll be unstored is when I'm dead and someone is sorting through my stuff. So it's only a natural step for me to want to be able to buy music legally without having to buy the physical CD and rip it to FLAC.

I don't see what new features in FLAC 2 would make the retailers sell FLAC files.

I'm sorry if my post was unclear.

I was meaning more the first question in the title of the post, the future of FLAC, rather than a FLAC 2.0. I think 2.0 has already been covered sufficiently in the thread to the point where it's reasonably clear it wouldn't be fruitful.

Absolutely; I already have. NIN's Ghosts I-IV project will be the model for many future digital releases, I'm sure...

I'm not a NIN fan at all, but I very much hope that launch went well for them. I'd like to see an iTunes type store that sells FLAC albums without DRM. I don't give a shit if I have to pay the price of a physical CD for them, I don't buy that many albums.

I want an archive on my PC of all my music in FLAC, all properly tagged with a cover image etc. And I'd like it DRM free so that I can encode it to a decent lossy format for my DAP.

It's always seemed daft that they want to give us lossy DRM-ed crap, when all it takes is firing the CD into the drive and we have a perfect lossless copy.

Personally even 20 or 30% compression is not enough. I just bought 4 1TB drives for 800 bucks and am running them in RAID 1 with 2500 CDs worth on them. That works out to 30 cents/cd.

As an aside the next big shift in computing will be the disappearance of the mechanical HD. Toshiba just announced 14 billion USD in spending on fab plants for flash memory. Imagine 1TB memory sticks. Take you entire music collection with you, in the car or on your portable. There are a couple of PC makers that have announced solid state HDs. You plug them into a HD interface but no moving parts. When memory comes that big and cheap compression really no longer matters.

I was talking about infinite decoder memory processing memory, not infinite storage or transmission bandwidth.

-k

Note, that recently there were reports on high return rate of notebooks with SSD (I originally wrote "SSDD" - and thought it sounds familiar. It's "Same shit, different day" from King's Dreamcatcher)

Query to Google Search hdd solid state high return rate

Why would you run 4 1TB drives in a RAID-1 arrangement? That would mean only 1TB of storage, essentially wasting 2TBs of redundant redundancy...

-brendan

While I agree that a stabilized FLAC 1x will most benefit "mainstreaming" FLAC more, I find the "we've reached the compression limits!" posts laughable.

- There is always a better way. Every year we find new ways of compressing data. 5-10 years from now, we may look at FLAC 1x and mock it's compression.

- CPU speeds and memory will continue to evolve. Period. That's just how it is. The beauty is that both compression AND hardware evolve. So things get smaller, and we can process more things at once. Good stuff.

Keep up the good work everyone. If you're working on stabilizing/standardizing in order to get FLAC 1x into more [commercial] products, you're doing the right thing. If you're thinking of better/faster/etc. compression methods, you're doing the right thing.

I can't see it being mentioned here, maybe it has, but speed improvements are always welcome!

I dont know where you got that info from, but in the last 10 years, lossless compression improvements have constantly become more and more minor (and with lossless compression, i do not just mean audio but the entire computer industry). Even lossy compression seems to be near its limits - not as much as lossless, but there isn't much on the horizon. Significant compression improvements, are no longer probably, unless some kind of revolutionary invention happens. And revolutionary inventions, aren't an improvement of something existing, but a complete rethought.

...and yet developers keep chipping away, getting ever nearer to whatever natural limit exists (though I agree, I think we're nearing this limit).

Places where we're seeing significant improvement is in compression and decompression times.

Is there a new minor version of FLAC being worked on or?

there was but it's stalled right now. the only major new thing slated is rf64/wave64 support.

Having frame size in each frame header would be very nice for FLAC. Of course FLAC can be fitted in a container (ie. mkv) but I guess most people use native FLAC.

What's the most important for a codec?

My answer may not same as others, I choose "popular support".

Just download a Foobar2000, FLACdrop (minor problem) and install, then I can convert the format easily, without painful parameters, for native playing through media player just install OggCodec (outdated though), or download a free Cog for my Mac, it's convenient.

FLAC is now popular (of course not for the popular brand such as Philips or Panasonic), it's adopted in quite a lot hardware, format is stable enough, maybe how to make it more efficient is the next step.

We're getting more cores from the CPU, maybe it's good time to consider tweak the encoding or decoding process use less CPU or go SMP, or port the encoder/decoder to a better OS such as OSX or Linux.

I'm quite sure the future Windows environment will limit further code improvement.
Vista? Windows 7? Thanks but no, I'm too tired.

Thanks Josh.

FLAC is already cross-platform, and most linux distros include FLAC tools/libs out-of-the box.

Multi core FLAC encoding could be useful for people encoding whole CDs - but anybody encoding tracks can just run them in parallel anyways. For decoding, the operation is so cheap on a PC CPU that there is little point. FLAC has had command line tools (and player support) on OSX and Linux for a long time.