Hydrogenaudio Forums

Lossless Audio Compression => FLAC => Topic started by: IgorC on 03 November, 2007, 01:28:00 AM

Poll
Question: FLAC
Option 1: -0 votes: 3
Option 2: -1 votes: 1
Option 3: -2 votes: 1
Option 4: -3 votes: 2
Option 5: -4 votes: 4
Option 6: -5 votes: 102
Option 7: -6 votes: 55
Option 8: -7 votes: 4
Option 9: -8 votes: 298
Option 10: -8 -A tukey(0.5) -A flattop votes: 20
Title: FLAC -0 ... -8
Post by: IgorC on 03 November, 2007, 01:28:00 AM
What makes FLAC so widely used? Because it's asymmetric? Are people satisfied enough by compression ratio or is it price for high decoding speed?

Last time I start to compress my CD rips with even more asymmetric settings:
-l 12 -b 1152 -r 6 -e

Compression ratio is as of -4 preset but decode speed is higher than for -3 and 10% slower  than -0.  It's of course at cost of lower encoding speed than of -4 preset.
Title: FLAC -0 ... -8
Post by: Synthetic Soul on 03 November, 2007, 06:14:50 AM
It is interesting to see, so far, that the compression level is paramount with the majority of users.

I did not vote, as I don't use FLAC; however, if I did: I would use -6, as it seems like the best compromise between compression and speed in my tests, and some others.
Title: FLAC -0 ... -8
Post by: guruboolez on 03 November, 2007, 06:30:30 AM
-8
On my side encoding speed is perfect for CD Ripping (x20, which is already faster than my drives in secure mode) and also decent for batch encoding (x40 with two cores).
Title: FLAC -0 ... -8
Post by: Mitch A on 03 November, 2007, 06:37:22 AM
-5, I've always just used this 9 (Being the default). It's considerably quicker than -8 and there barely any difference in file size
Title: FLAC -0 ... -8
Post by: user on 03 November, 2007, 07:56:10 AM
What makes FLAC so widely used? Because it's asymmetric? Are people satisfied enough by compression ratio or is it price for high decoding speed?


Personally I use -8 , which is even fast enough for old Pentium 3 @ 800 MHz, maybe as I rip EAC , Test & copy secure.

IIRC, decoding speed of flac is more or less always the same very fast decoding speed, independent of encoding setting, ie. once you compress to the max with slightly more encoding time, you get the flac smalles sizes, and have also the very fast decoding.

In general, I think flac is so widely used because:
- very fast encoding, if you don't use -8
- more important: the als very very fast DEcoding.
- hardware support by industry , flac was the first (free) lossless codec, I think
- same effect like mp3, as flac has the majority, it grows even quicker, though other lossless formats might have slight advantages on certain single aspects,
but flac is strong at: en- and maybe more important DEcoding speed, hardware support.
- Compression ratio differences amongst lossless codecs are slight and practically unimportant.
What might matter to end-users: speed & hardware/software/OS support.
flac was iirc the first lossless codec for all major OS like windows, unix/linux, mac ?
Title: FLAC -0 ... -8
Post by: drbeachboy on 03 November, 2007, 11:21:11 AM
I use -8 now, though I was using -6 for a very long time. My 320G HD is starting to fill up fast, so I need all the extra compression I can get. Plus, versions 1.2.0 & 1.2.1 are much faster, so encoding time is not as big a deal as it used to be.
Title: FLAC -0 ... -8
Post by: Clear89 on 03 November, 2007, 11:34:05 AM
I use -8 now, though I was using -6 for a very long time. My 320G HD is starting to fill up fast, so I need all the extra compression I can get. Plus, versions 1.2.0 & 1.2.1 are much faster, so encoding time is not as big a deal as it used to be.


I second that. The encoding speeds just get faster and faster with each new release, so the wait is being phased out.

-8 all the way.
Title: FLAC -0 ... -8
Post by: Borisz on 03 November, 2007, 11:41:46 AM
I never understood, what does -A tukey(0.5) -A flattop do?

Also, -8, no point in using anything less.
Title: FLAC -0 ... -8
Post by: verbajim on 03 November, 2007, 01:53:17 PM
I've gone the other way, from 8 to 6 because I realized the compression loss was too small for me to care about. The speed gain is much greater.
Title: FLAC -0 ... -8
Post by: beto on 03 November, 2007, 02:42:00 PM
-5 for me. I used to use -8 but since there is no significant compression advantage for me I can use the extra speed in compressing.
Title: FLAC -0 ... -8
Post by: lunkhead on 03 November, 2007, 02:57:20 PM
5, but maybe i'll try 6.

did you have to include the dashes?
Title: FLAC -0 ... -8
Post by: Synthetic Soul on 03 November, 2007, 02:57:31 PM
flac was the first (free) lossless codec, I think
A bit before my time (here) but wouldn't Shorten  be better suited to hold that accolade?  There may be even earlier codecs.

I never understood, what does -A tukey(0.5) -A flattop do?
IIRC recent versions of FLAC (1.1.3 + I think) use -A tukey(0.5) by default in the higher compression settings.  Using two -A switches (http://flac.sourceforge.net/documentation_tools_flac.html#flac_options_apodization) will increase compression ever-so-slightly, but will also drastically reduce encoding speed.  I really can't see the point.
Title: FLAC -0 ... -8
Post by: Engywuck on 05 November, 2007, 04:19:04 AM
I usually use -8, since even when ripping 2 CDs simultanuously I seldom have even one encoding running, my PC being fast enough to encode MUCH faster even in -8 than EAC is with testing and ripping in secure mode. So why spend more hard disk space than necessary?
Title: FLAC -0 ... -8
Post by: Bad Monkey on 05 November, 2007, 08:41:11 AM
I usually use -8, since even when ripping 2 CDs simultanuously I seldom have even one encoding running, my PC being fast enough to encode MUCH faster even in -8 than EAC is with testing and ripping in secure mode. So why spend more hard disk space than necessary?

I don't get this, EAC for me always rips the CD to a WAV then pauses while FLAC encodes. This is annoying because I used to use AudioGrabber which encoded the MP3 while ripping, instead of just having the CPU practically idle then the CD drive idle and so on. Anyway it means the speed of the FLAC encode stage is relevent, otherwise obviously it only really needs to be as fast as the rip in practical terms.

Is there a way of getting EAC to encode direct without the intermediate WAV?
Title: FLAC -0 ... -8
Post by: M on 05 November, 2007, 09:01:49 AM
What makes FLAC so widely used? Because it's asymmetric? Are people satisfied enough by compression ratio or is it price for high decoding speed?

Perhaps one of the most significant factors in FLAC's public adoption was the embrace of the taper-friendly touring band community. Shorten was, as Synthetic Soul intimated, originally the choice for trading such material, but as FLAC achieved superior speed and compression ratios - and given it's cross-platform compatibility (another big factor for live audio archivists) - it soon supplanted Shorten as the codec of choice.

Arguably, Monkey's Audio was at one point vying for the same status... but given the number of systems on which the decoding of APE files was simply too processor-intensive to be feasible on a large scale, it is (mostly) only among the fans of certain bands that Monkey's Audio is still considered a desirable trade format.

    - M.
Title: FLAC -0 ... -8
Post by: Borisz on 05 November, 2007, 12:28:58 PM

What makes FLAC so widely used? Because it's asymmetric? Are people satisfied enough by compression ratio or is it price for high decoding speed?

Perhaps one of the most significant factors in FLAC's public adoption was the embrace of the taper-friendly touring band community. Shorten was, as Synthetic Soul intimated, originally the choice for trading such material, but as FLAC achieved superior speed and compression ratios - and given it's cross-platform compatibility (another big factor for live audio archivists) - it soon supplanted Shorten as the codec of choice.

Arguably, Monkey's Audio was at one point vying for the same status... but given the number of systems on which the decoding of APE files was simply too processor-intensive to be feasible on a large scale, it is (mostly) only among the fans of certain bands that Monkey's Audio is still considered a desirable trade format.

    - M.

Also, p2p.
Oink recommended, and made proper guides for, ripping in flac. With tens of thousands of rippers, thats a factor.
APE+cue cd image are still widely used in edonkey and japanese p2p (winny, share). On japanese systems however, TTA+cue cd images are starting to spread as well, mainly because True Audio is developed by a japanese person I believe.

Also flac was heavily promoted by xiph.org, and now has limited hardware support, which is another factor, since people will just associate lossless audio with flac in general (despite the best tries of both Apple and Microsoft with their own lossless formats).

Ah, the golden old days, when I used ape/cue myself for archiving, and Musepack for lossy copies...
Title: FLAC -0 ... -8
Post by: verbajim on 05 November, 2007, 02:26:47 PM
I don't get this, EAC for me always rips the CD to a WAV then pauses while FLAC encodes. This is annoying because I used to use AudioGrabber which encoded the MP3 while ripping, instead of just having the CPU practically idle then the CD drive idle and so on. Anyway it means the speed of the FLAC encode stage is relevent, otherwise obviously it only really needs to be as fast as the rip in practical terms.

You can make EAC encode in the background while ripping if you go to EAC options, under tools, and check "On extraction, start external processors queued in the background".
Title: FLAC -0 ... -8
Post by: Seiitsu on 05 November, 2007, 02:30:47 PM
-8

It encodes really really fast for me... so no reason to use anything less.
Title: FLAC -0 ... -8
Post by: jcoalson on 05 November, 2007, 02:34:48 PM
flac was the first (free) lossless codec, I think
A bit before my time (here) but wouldn't Shorten  be better suited to hold that accolade?  There may be even earlier codecs.
depends on what's meant by 'free'.  shorten was only free for non-commecial use.  ogg squish was also around before flac but not actively developed.
Title: FLAC -0 ... -8
Post by: GeSomeone on 05 November, 2007, 04:52:33 PM
depends on what's meant by 'free'.  shorten was only free for non-commecial use.  ogg squish was also around before flac but not actively developed.

I never had to pay for LPAC and RKAU had also a lossy mode. But in that time lossless files were much too big for my (small) disks.

BTW only -8 when I use FLAC
Title: FLAC -0 ... -8
Post by: probedb on 08 November, 2007, 03:24:18 AM
-8

It encodes really really fast for me... so no reason to use anything less.


Ditto  Not like I'm in a rush to encode either...
Title: FLAC -0 ... -8
Post by: Bad Monkey on 08 November, 2007, 03:27:47 AM

I don't get this, EAC for me always rips the CD to a WAV then pauses while FLAC encodes. This is annoying because I used to use AudioGrabber which encoded the MP3 while ripping, instead of just having the CPU practically idle then the CD drive idle and so on. Anyway it means the speed of the FLAC encode stage is relevent, otherwise obviously it only really needs to be as fast as the rip in practical terms.

You can make EAC encode in the background while ripping if you go to EAC options, under tools, and check "On extraction, start external processors queued in the background".

No way to pipe it straight to the encoder while it rips, then?
Still a lag and wasted time as it encodes the last track, if I understand this right.
Title: FLAC -0 ... -8
Post by: ArtMustHurt on 10 November, 2007, 12:19:44 PM
-8 here as well
Title: FLAC -0 ... -8
Post by: collector on 17 November, 2007, 07:58:14 AM
In the longhelp I read that -8 and -7 have -e embedded. Exhaustive-model-search (expensive!), for what it's worth. So I take -6 as a default. It's fast enough and -8 and -7 are hardly better in compression; also harddiskspace isn't an issue for me.
Title: FLAC -0 ... -8
Post by: EagleScout1998 on 17 November, 2007, 05:37:20 PM
No way to pipe it straight to the encoder while it rips, then?
Still a lag and wasted time as it encodes the last track, if I understand this right.


Are you talking about "on the fly" encoding (like what CDex does)?
Title: FLAC -0 ... -8
Post by: Bourne on 17 November, 2007, 05:47:43 PM
I hope Blue-Ray/HD-DVD Players come with FLAC support.
Title: FLAC -0 ... -8
Post by: IgorC on 17 November, 2007, 06:08:42 PM
I hope Blue-Ray/HD-DVD Players come with FLAC support.

Maybe some chipmakers will go for FLAC but that won't be defined by standarts.  There is  Dolby True HD for lossless purpose.
Title: FLAC -0 ... -8
Post by: djdust on 02 December, 2007, 11:48:00 AM
Complete newbie thickie here so, apologies if this is a dumb question..  Can someone explain what the difference is with the 0-8 settings?  I just took a WAV file and tried compressing it with each setting and found that "0" produced the largest file size and "8" the lowest.  I thought that "8" was the best so I would have thought it would be the largest!?  *confused*
Title: FLAC -0 ... -8
Post by: SamHain86 on 02 December, 2007, 11:52:33 AM
@ DJdust:
The 0-8 is the compression setting. With FLAC there is no actual best, since all 0-8 settings are lossless, there is no audio data taken away. The only measurement then is the amount of compression that FLAC encoder can apply to the WAV. 0 being the least compression, 8 the most compression. Therefore, 0 has the largest files, 8 has the smallest files for FLAC. With compression there is usually a sacrifice of time. You probably noticed that the 0 setting took the least time, and the 8 setting took the longest.
Title: FLAC -0 ... -8
Post by: djdust on 02 December, 2007, 12:57:00 PM
@ DJdust:
The 0-8 is the compression setting. With FLAC there is no actual best, since all 0-8 settings are lossless, there is no audio data taken away. The only measurement then is the amount of compression that FLAC encoder can apply to the WAV. 0 being the least compression, 8 the most compression. Therefore, 0 has the largest files, 8 has the smallest files for FLAC. With compression there is usually a sacrifice of time. You probably noticed that the 0 setting took the least time, and the 8 setting took the longest.

Actually, it seems to take the exact same amount of time really so, I guess I'll stick with setting "8" then and get the smallest file size! 
Title: FLAC -0 ... -8
Post by: damaki on 02 December, 2007, 01:49:01 PM
@ DJdust:
The 0-8 is the compression setting. With FLAC there is no actual best, since all 0-8 settings are lossless, there is no audio data taken away. The only measurement then is the amount of compression that FLAC encoder can apply to the WAV. 0 being the least compression, 8 the most compression. Therefore, 0 has the largest files, 8 has the smallest files for FLAC. With compression there is usually a sacrifice of time. You probably noticed that the 0 setting took the least time, and the 8 setting took the longest.

Doesn't -0 sounds warmer and fuller?
Title: FLAC -0 ... -8
Post by: SamHain86 on 02 December, 2007, 02:02:16 PM
@ Damaki:
... AH HAHAHAHAHAHAHAHAHA!-
You nearly made me spit tea all over my damned laptop.

@ DJdust:
I am with you about using -8, same as --best. Just take "my" word (http://flac.sourceforge.net/comparison.html) that 8 takes longer than 0. I encode to FLAC images at --best and I do not care about the additional time.
Title: FLAC -0 ... -8
Post by: miranon on 06 February, 2008, 01:55:46 PM
I use -6. Best compromise for me between 7-8 and 5.
Title: FLAC -0 ... -8
Post by: naturfreak on 06 February, 2008, 02:48:14 PM
Default setting (-5), I don't care about 0.5% better compression of -8
Title: FLAC -0 ... -8
Post by: ArtMustHurt on 11 March, 2008, 08:09:45 AM
i use -6, i ripped a bunch of cds last night and re-encoded some old flacs with the 1.2.1 and damn its pretty fast to encode now
Title: FLAC -0 ... -8
Post by: 2tec on 20 April, 2008, 10:59:25 AM
Using -8 myself, why compromise? Besides with multicore systems, encoding time is a moot point. No?

As well, how come no poll option for -totally_impractical? 
Title: FLAC -0 ... -8
Post by: Soap on 05 April, 2009, 04:16:14 PM
-8.
Even if I only got 0.01% better compression than -3 or -5 I'd still do it.  It isn't like I'm doing the math with a slide rule.
Title: FLAC -0 ... -8
Post by: KFal on 06 April, 2009, 05:26:05 AM
When I ripped the entire CD collection of my girl friend and mine, I used -5 because at the time there was a significant speed difference to -8 and disk space was of less concern. I had only second hand comparisons for the decoding speed for a Squeezebox but it did not appear to have any perceptible difference in the starting of songs.

So, without much further experimentation I chose the default.
Title: FLAC -0 ... -8
Post by: insane_alien on 06 April, 2009, 08:40:28 AM
-8

it shaves a few gigabytes off my collection and the encoding times do not bother me at all as i only have to do it once(well, maybe a second time if a new version of FLAC comes out that makes my lossless music even smaller.). decoding times are also a non-issue as it only needs to be something other than realtime once in a blue moon when 5 austrailian aborigional mimes are pretending to be in an invisible box on the top floor of the eiffel tower.
Title: FLAC -0 ... -8
Post by: Angenial on 06 April, 2009, 01:08:29 PM
What makes FLAC so widely used? Because it's asymmetric? Are people satisfied enough by compression ratio or is it price for high decoding speed?

For me it's the fact that it's open, and that's not just an ideological reason.  Since I use Linux and FreeBSD, the programs I use tend to be open, as well; and that means that if a file format is closed, odds are I won't be able to use it well.  Perhaps the ffmpeg team will wind up implementing reverse-engineered versions (as happened with ALAC), but why bother with that when flac and associated tools (like metaflac) are available?

If something open came along with a much better compression ratio, perhaps I'd switch; but then again, all my CDs are ripped and fit just fine on my 500GB array (with room to spare), so it probably wouldn't be worth the bother.
Title: FLAC -0 ... -8
Post by: madoka@ex-sounds on 06 April, 2009, 01:30:28 PM
-8

for peace of my mind 
Title: FLAC -0 ... -8
Post by: _m²_ on 07 April, 2009, 03:25:17 PM
-8.
Considering that I don't plan recoding more often than once / few years, encoding time is negligible while space taken is not. -8 -A tukey(0.5) -A flattop is stronger? Got to make a switch.
Title: FLAC -0 ... -8
Post by: tev777 on 07 April, 2009, 10:53:32 PM
--best here. Why use any other setting when you can have the "best"? Or is Josh just using Jedi mind tricks on me?

As for using FLAC, I switched to it when I started using Linux because it was supported out of the box. It has served me well over the years and I see no reason to switch to anything else. I am really impressed by Wavpack and enjoy watching it grow, but FLAC FTW!

Title: FLAC -0 ... -8
Post by: ottl on 08 April, 2009, 05:41:20 AM
-8. As many, I think the extra wait is no problem on a modern dual/quad core machine. I rip and transcode with XLD (FLAC 1.2.1) and it works quite smoothly.
Title: FLAC -0 ... -8
Post by: sshd on 08 April, 2009, 06:29:32 AM
-5 is a good default choice.

-8 is a waste of time saving very little space. I only use it for mkv movies with eac3to as it is the only option.
Title: FLAC -0 ... -8
Post by: ktf on 13 April, 2009, 01:42:29 PM
-8 -A tukey(0.5) -A flattop is stronger? Got to make a switch.


I tested it on a song and it is actually worse for that song, but less then 1/1000 (it just shows ratio=1,000) I've played around with those -A switched and other settings under --lax, but the gains are extremely small and encoding takes ages xD

I use -8, just because it is --best    Indeed, it is probably not the worth the slowdown, but I don't care. New versions usually make much bigger differences.
Title: FLAC -0 ... -8
Post by: JJZolx on 19 April, 2009, 09:52:12 PM
Less than zero.   

--max-lpc-order=0 --disable-constant-subframes --disable-fixed-subframes
Title: FLAC -0 ... -8
Post by: frozenspeed on 19 April, 2009, 10:31:46 PM
What the heck is flattop?
Title: FLAC -0 ... -8
Post by: Justin Ruggles on 20 April, 2009, 09:37:34 AM
What the heck is flattop?

http://en.wikipedia.org/wiki/Window_functi...ow_.5Bnote_1.5D (http://en.wikipedia.org/wiki/Window_function#Flat_top_window_.5Bnote_1.5D)
Title: FLAC -0 ... -8
Post by: hazumi-san on 24 April, 2009, 10:45:10 AM
I just started using flac and I used the -8....
encoding or decoding speed is not a problem as long as I get the best result.
actually the speed is quite fast with my pentium D desktop.... (encoding and decoding)
and the filesize produced is not really big to me, much smaller than wav, which I used to back up
my song library before.

Title: FLAC -0 ... -8
Post by: ManekiNeko on 03 May, 2009, 10:36:10 AM
Flac -8 makes virtually no difference in compression/filesize over -6, but does take over twice as long to encode. If space saving is your main objective, flac is not the right choice. If very fast decoding and compatibility is, then use flac.
Title: FLAC -0 ... -8
Post by: acetylcholine on 05 May, 2009, 05:10:25 AM
-8 for me.

Using a multi-core processor and it takes so much longer to rip than to encode so there is no issue about encoding duration.
Title: FLAC -0 ... -8
Post by: hero on 11 May, 2009, 03:42:13 PM
The final size doesn't really differ between the compression options. I get about the same speed with -8 than I would with -1. When choosing the compression level, I guess it all depends on your system.
Title: FLAC -0 ... -8
Post by: doccolinni on 29 May, 2009, 07:18:19 AM
I use -8. I really don't have any trouble at all with waiting a bit longer to achieve the best compression possible. Of course, that doesn't mean that I'd use a coder which takes half an hour to encode a song. :-/ I just don't think FLAC is really all that slow when I use compression level of -8. Also, I could never force myself to use FLAC with a lower compression level and use those files for storage when I'd know that I can bring their size down even further, no matter how little difference it would be.

So yeah, -8 all the way.
Title: FLAC -0 ... -8
Post by: collector on 30 May, 2009, 06:43:29 AM
-6, seems to be the best of both worlds. And I'm uing lossywav -q6, so why bother about size or speed. 
Title: FLAC -0 ... -8
Post by: jmcguckin on 22 June, 2009, 12:11:30 PM
I've been using -8 for several years now, only because I'd like to get the smallest filesizes possible, and now that I have a dual-core processor at my disposal, it doesn't make sense to use anything less... I use Max for ripping/encoding, and encoding speeds are typically somewhere between 1.5-2x faster than ripping speeds, encode speed is pretty much a non-issue.