Initially, let me apologize if this has been posted in the wrong forum. I took a look around, and it appeared that over all this was most likely the most appropriate place for this.

Secondly, on one hand I feel like this is a newb question that could probably be answer by searching through old topics, but in all honesty I've spent the last hour browsing or so, and have not come up with an exact answer.


My basic situation is this: I would like to back up my entire CD catalogue (approx 800 CD's as of right now) into a digital format on my my hard drive. I am doing this as much as to have a permanent back up of my music as for ease of listening while I'm at my computer.

I would like to use a compressed format, because using something lossless would obviously eat up so much room it would be ridiculous (or not? Am I wrong here? This is why I'm posting this question in the first place!)

I would prefer to use a format that can be asured to be easily supported for the forseeable future (is Vorbis better than MP3 in this regard?)

I am a bit of an audiophile, and a high bitrate is most desired. What is the best ratio of compression -> quality? Which codec? I would like to retain as much quality as possible, but can't afford 700megs per CD. What is the generally accepted consensus on where quality and compression meet?

I run Linux systems.

Guess that's about it... any input would be greaty appreciated! If you have further questions about my setup or if I didn't make the requirements clear, just let me know...

Thanks in advance!!!

-Alex

Lossless can still be compressed; you're thinking of lossily compressed audio. Losslessly compressed audio does take up more room than lossy at conventional bitrates. Usually, lossless compression can provide bit-for-bit replication with about 900kbps audio (of course, this depends on the music being encoded), which is considerably less than 1411kbps for uncompressed PCM. If you are getting rid of your CDs, or backing up a lot (which you definately are in this case), I suggest you go this route.

However, a great compromising option would actually be to use WavPack. Using WavPack's hybrid mode, you get a very good lossy file (that can compete with MusePack) AND a correction file. When these two files are combined, the can be decoded to a bit-for-bit copy of the origional. This is particularly good for storing high-quality lossy audio on you're computer, and have the ability of recovering the same file somewhere offsite.


Continuing with the WavPack reccomendation, it is likely that this format will be supported for a long time. It is
open-source, so can be used on practically any system (of course including Linux).


WavPack is appx. perceptually lossless (IMO) at about the same bitrates as MusePack (~200kbps). This produces 6-8MB files for average songs. If you do have the option for CD/DVD burning, though, you can get the recovery files from hybrid mode off of your computer and still have high-quality ones on your computer.

However, all this regards WavPack. There are many other good formats, but for your particular need (archiving MANY CDs), WavPack hybrid mode at ~200kbps sounds good to me.

EDIT: Here's a good thread for lossy(/hybrid) WavPack info.

EDIT2: Reading more, I think WavPack ~300kbps is appx equal to MusePack ~>200kbps. I haven't tested WavPack as in depth as I have MusePack.

EDIT3: You're running linux. I don't know why I put dbPowerAmp up there. Also, I'm not sure if there's a linux port of WavPack yet, but maybe you can find/make a compile.

One word: FLAC.

It is an open source lossless audio compressor. Open source implies support on Linux, and also future support. Lossless means no quality loss whatsoever, every bit is preserved (kinda like zip, but with dedicated audio algorithms). This reduces files to roughly 60% of original size.

You can always transcode from lossless into any other format (e.g. for a portable or if a newer hotter format is available) without quality loss, whereas repeated lossy encodings drastically reduce quality. This is why lossless codecs are generally recommended for archival use.

If I make the generous assumption that your CDs are one hour on average, it's under 400 MB per CD. That's 320 GB for your collection. It is on the big side, but it really pays. I'd even suggest that you spread out your encoding over time, starting with the most precious/often used discs, so that by the time you fill up your existing HD the drives are cheaper and it's time to upgrade anyway.

If you really definitely cannot go lossless, you have a number of options. MP3 is the most compatible and therefore future proof. MPC is most transparent, but more exotic, hence less future-proof. Vorbis beats MP3 in lower bitrates but is not currently recommended for high bitrates. AAC is somewhat similar to MPC and gets broad support due to its use in MPEG 4. Once again, transcoding from lossy codecs is not recommended.

Finally, the software. Exact Audio Copy is believed to give most secure rips. It is Windows-only, but is reported to work well under Wine. If your CDs and the drive are well-maintained, you may be equally happy with cdparanoia.

P. S.: pseudo beat me. FLAC *does* exist under Linux, however does not sport a hybrid mode. MusePack = MPC.

FYI, 800 CDs will require about 240-50 GB of drive space using FLAC, at least that's how much space for the 800 pop CDs I've ripped so far (Exact Audio Copy + FLAC set to standard compression ratio).

I ripped them to 1 maxtor 250 GB drive, then backed that up to a WD 250 GB drive I then removed from the PC and set aside. It would be cheaper to use multiple smaller drives though.

The drive is in a Windows 2000 development server / music server PC (tualatin celeron 1.3a, 512 MB RAM, Asus 50x, Windows 2000) on a Promise 133TX2 controller along with a second drive I'll be using for my celtic / jazz / blues / classical CDs.

Time to extract and encode with EAC + FLAC is about 10 minutes per CD, and I have it set to eject when done so I can do something else at the same time without watching it.

I use Foobar2000 to play the FLACs as-is, no need to decode or re-encode as MP3s. For my flash mp3 player I have to decode / re-encode but the quality is exactly like going from CD to MP3, unlike starting with a high-bitrate lossy format. Also, SonicBlue is coming out with new Rio players with native FLAC support.

So far it sounds like FLAC may be the way to go. I like what I'm hearing about it, and what I've found out on my own by checking out the project page.

Currently I only have a seperate 60gig ext3 partition set up to rip to, but that will be upgraded as soon as possible. I'm a broke college kid, but I should be able to scrape together a few hundred dollars for a bigger HD soon. (120gig right now)

Those size numbers sound fine to me: although I will eventually want EVERYTHING backed up, there are obviously CD's in my collection more important to me than others, and CD's I am more apt to listen to than others. I don't have a time-frame to get this done or anything, as it doesn't really serve any REAL purpose other than assuring me I have my music backed up in a (fairly) secure manor. HD's don't scratch, and aren't lent to friends who never give them back. (Well, not NORMALLY )


New additional question: What about my drive? It's an old HP 9300+, probably about 4 or 5 years old now? It's, infact, the only part of my current system that is more than about a year old. Am I going to notice a lack of quality using this drive? I'm not having any problems with it, it's just "ancient" as computer parts go.

-Alex

Probably shouldn't be. Quality really isn't dictated by the drive so much as the program. If you can get cdparanoia to work nicely, then there shouldn't be a problem. The biggest problems are CD jitter and CD scratches, both of which are compensated for in both cdparanoia and Exact Audio Copy.

At worst, you'll have the same problem I have with cdparanoia, which is that it's really f***ing slow on a heavily scratched CD.

I backup my FLACs to DVD-R. I average 10-14 cds per DVD-R.