Like other people, I'm starting to think about ripping all my CDs into FLAC format. Although I haven't got a decent player yet, I'm starting to look at things like the SB+ (I've got a Squeezebox Boom).

Obviously I don't want to be ripping my CDs again so I want to get it right first time. As Linux user I used rubyripper (http://code.google.com/p/rubyripper/). However if I run this on two different machines the FLAC files generated are not only different but a slightly different size and this was using the same encoder. Rubyripper essentially does the actual rip twice to ensure integrity. My understanding of FLAC files was that as they are lossless and assuming that the bytes read from the CD are correct how is it they can vary?

My apologies if this is a FAQ but I couldn't find anything about it.

Hi,

Cannot speak as a Linux user but as a PC user I use dBpoweramp and it does not matter what PC I rip them on it uses the AccurateRip Data base and the files always come out the same. Even files ripped on other peoples PC's have the same size.

http://www.dbpoweramp.com/

Cheers

John

flac encoding can be adjustable to vary the file size a bit like a zip file.

have you set the tagging information to be the same on both machines? It may be that variations in tagging lead to different file sizes. (Although I don't know, this is a guess)

the 'music' component of the flac has to be the same assuming the ripping s/w is conforming to the standard. The various ripping programs will add their own metadata (e.g. tag placeholders, tags, version history...whatever).

Or are you saying they sound audibly different? Which would imply you should stop using one of your ripping apps as it's not doing its job!

Rob

There is different compression levels 0 - 8 that can be applied to FLACs which affect the final size. The general consensus is that it dosent affect the size or sound quality that much so go with the default

Other than embedded tags and the encoding level which have already been mentioned, CD Drives vary slightly as to where they start and stop reading data, so you can get slight differences in the length of the run-in and run-out. The FLAC encoding algorithm has changed over time too, so although there is always backwards compatibility and losslessness, encoding efficiency might be slightly different from version to version.

It is well known that encoding efficiency can vary, while remaining lossless and compatible with the same decoder.

To take a simple example, think of run length encoding, which compresses a repeated value by outputting a repeat count and the value. An implementation that can count up to 4 might transform a group of 12 zeros into:
4x0 4x0 4x0
A smarter implementation might count up to 8, giving:
8x0 4x0
Still better, you might get:
12x0

All versions are lossless, all can be reconstructed by the same decoder, but they are different lengths. The example is simplistic, and real audio encoders are much smarter than RLE, but you get the idea.

Thanks for all the replies. Upon reflection the encoders are slightly different versions so as been pointed out that may well account for it.

I can't say I could detect any different in sound in the tracks, but it sounds like I probably don't need to get worried about the file sizes differences.

Interesting link to the accurate rip database. As a developer myself perhaps I should see if I can get access to their API to contribute.

Thanks again guys (and possibly gals!)

http://www.designwsound.com/dwsblog/?page_id=535