FLAC - Speed & Comression factor

[tuesday]

Hmmm ... a bit confused since I thought that as long as i compressed in FLAC , there where no really options in this format (as there are with MP3). But when I started to compress today, using Winamp 5.531, I had several options. What do they really do?

Speed: choose from 0.5 - 8
What happen if I crank this one up as much as possible?

Compression factor: choose fast encoding --- best compression
Ceartinly I choose best compression, but why are ther more options?

Can anyone tell me what to do.

Thanks guys!

 

[krmathis]

Quite simple.

0.5 = Faster encoding and larger files.
8 = Slower encoding and smaller files.

There are options because we have different requirements and preferences. Some want the maximum compression regardless of encoding time, while others want fast encoding time and don't care that much about file size.

 

[xenithon]

Yup....just to add to krmathis's explanation, no matter what speed/compression factor you choose, FLAC remains lossless (unlike lossy MP3).

The levels of compressions will determine a) time it takes to compress/decompress; b) the amount of disk space it takes up; and c) the CPU load needed for compr/decompr.

Cheers,
X

 

[tuesday]

Quote:

Originally Posted by krmathis /img/forum/go_quote.gif Quite simple. 0.5 = Faster encoding and larger files. 8 = Slower encoding and smaller files.

Ok, but (sorry for my ignorance) in winamp
it says 0.5 x speed -> 8 x speed (while over 8 you have to buy winamp pro

). That must mean that 8 is faster

Ok xenithon got at least that!

Cheers!

Now I´ve tried one track at different speeds (1 & 8):
-No difference in size, no significant difference in time ....

 

[LeonvB]

Quote:

a) time it takes to compress/decompress; b) the amount of disk space it takes up; and c) the CPU load needed for compr/decompr

A and C aren't true for the decoding: FLAC searches for functions to replace the signal with a perfectly matching formula to re-calculate the signal at runtime. The higher setting means the FLAC encoder searches longer for a match, deviations when decoding are negliable most of the time.

As a grossly simplified example, consider these calculations:
100-97
10-7
3-0
Of course the last one takes the least amount of space to represent an outcome of 3. By selecting the 8 setting FLAC simply tries longer to find the best formula, but that does not mean the calculation needed at runtime actually takes longer to generate the signal.

 

[krmathis]

Quote:

Originally Posted by tuesday /img/forum/go_quote.gif Ok, but (sorry for my ignorance) in winamp it says 0.5 x speed -> 8 x speed (while over 8 you have to buy winamp pro ). That must mean that 8 is faster

I don't use WinAmp (or MS Windows ...), so don't know what kind of settings it use and how the rely to the FLAC settings.

But for the command line FLAC encoder the settings give the result I mentioned above.


Edit.
I performed a short test and get these results (original file 67130176 bytes):
Quote:

$ time flac -0 Angel.wav <snip> Angel.wav: wrote 41214627 bytes, ratio=0.614 real0m2.555s

Quote:

$ time flac -8 Angel.wav <snip> Angel.wav: wrote 37409700 bytes, ratio=0.557 real0m13.628s

 

[jilgiljongiljing]

Also I think there was some issue with some players not supporting compression levels more than 5. I think I read that on another thread here a while ago.

But then since I only use Foobar, I just stick to 8 for smallest file size and I havent had any issues.

 

[infinitesymphony]

Just install the latest version of FLAC using the latest FLAC for Windows Installer, then use the included FLAC Frontend. Drag-and-drop, select your options, and hit Encode.

 

[indikator]

a little OT,

why don't you guys use uncompressed instead?
large hard disk has become really cheap nowadays right?
so close to perfection, why not go for it instead?

 

[infinitesymphony]

Quote:

Originally Posted by indikator /img/forum/go_quote.gif a little OT, why don't you guys use uncompressed instead? large hard disk has become really cheap nowadays right? so close to perfection, why not go for it instead?

Because when it's compressed, you can fit more music on the same drive with no loss in quality. There are no benefits to uncompressed files, unless you're editing them or burning CDs.

 

[krmathis]

Quote:

Originally Posted by indikator /img/forum/go_quote.gif a little OT, why don't you guys use uncompressed instead? large hard disk has become really cheap nowadays right? so close to perfection, why not go for it instead?

Simply because its a waste of space.
Lossless encoded audio have an average file size of 65-60% of the uncompressed source file. With the additional benefits that you get full tag and album art support (unlike WAV)

 

[tuesday]

Quote:

Originally Posted by infinitesymphony /img/forum/go_quote.gif Just install the latest version of FLAC using the latest FLAC for Windows Installer, then use the included FLAC Frontend. Drag-and-drop, select your options, and hit Encode.

Yeah! Looks great! Can you encode directly from a CD as a source or do you have to make, type WAV files first?

 

[progo]

Quote:

Originally Posted by tuesday /img/forum/go_quote.gif Yeah! Looks great! Can you encode directly from a CD as a source or do you have to make, type WAV files first?

Having used that frontend 10 months ago, it ate only waves, so you have to rip the CDs to wavs first. EAC does that handily.

 

[Apocalypsee]

Quote:

Originally Posted by krmathis /img/forum/go_quote.gif Simply because its a waste of space. Lossless encoded audio have an average file size of 65-60% of the uncompressed source file. With the additional benefits that you get full tag and album art support (unlike WAV)

Well said. Although some album I do use WAV

I use FLAC level 5

 

[krmathis]

Quote:

Originally Posted by tuesday /img/forum/go_quote.gif Yeah! Looks great! Can you encode directly from a CD as a source or do you have to make, type WAV files first?

No idea about the frontend.
But you can most probably set up your CD ripper application (whatever you use) to encode directly to FLAC using the command line binary.