All files need encoding. WAV is a form of encoding. The audio on the CD is encoded in a certain way. Phonograph records are encoded in a certain manner. Even the signals going over the wires to your speakers must be encoded in a certain way. And, as I'm not sure that you're completely following, FLAC introduces zero artifacts (barring bugs). It's like ZIP, except optimized for music. You get back out exactly what you put in, bit for bit.

What I'm wondering is why you're so interested in WAV files. I understand that it would take a while to compress them (which could potentially be obviated by encoding them directly to FLAC when ripping), but if you're not listening to them directly, why are you interested in having them? If your reason is that you might need to reencode them into mp3s (or WMAs or whatever), FLAC decompression still works much faster than mp3 compression, so if you just pipe the output of flac to the input of lame, since flac is presenting data faster than lame can accept it, you're not losing any time, other than, possibly, a few milliseconds per song for it to present its initial data. That is, the bottleneck would be lame, not flac.

Argh. I just bid on a Snap server because of this thread. Thanks guys

LoL... I was tempted to as well. See what you people do ... Actually what I want is my gateway (cobalt cube) microserver... it was really cheesy but with a few hacks the whole box was open to me. Too bad I had to to give it back to gateway after they let me try it out. I shoulda kept it till they discontinued it.. hehe

-Greg

I was just looking at Snap servers. Given that you can get a 160GB hard drives for about $300 and a bare bones system for about $100 (running a free-of-cost OS, be it Linux, FreeBSD, OpenBSD, Solaris, whatever), that means that you could get a 320GB of mirrored NAS for $1300. The Snap 4100 with 240GB retails for $3000 (Pricewatch minimum seems to be about $2500). Oops; I'm sorry, actually, that Snap server would only provide 120GB in a mirrored configuration. And the homebuilt one would be way more upgradable. So that's $4/GB versus $20/GB.

Hope you're getting a good deal at the auction.

I can't find a Linux box 1U in size with 120GB of capacity for (max bid of) $505.

That's 4x30GB Drives. If the hack pages are right, as time goes, I buy a new drive, swap, and I have more space.

1 yr warranty refurbished. Warranty is from Quantum.

Seemed Like a deal to me. Of course now that I've told everyone here about that I'm sure I'll get outbid. They only have 5 a tthat price.

$505 sounds like a good deal. Most folks don't actually have racks or cabinets at home, so I wasn't taking that into consideration. The homebrewed machine could have an assload of drives in it, though (but you might have to get a real kickass power supply). Think about it -- 4 drives on the motherboard, 4 more per PCI slot. Get a motherboard with enough PCI slots (and, again, a kickass power supply, or multiple ones) and you could have 24 drives in the machine, physical space allowing. Plus, it's easy to fix if it breaks. For this purpose, parts is parts.

This is the way I do it, and from free (or in some cases very cheap) second hand kit from 486's and ancient SGI Personal IRIS's through to my latest Athlon boxes I have a total of 1.1ish Terabytes available for use. I have all of it shared using Samba and NFS and it works beautifully. Admittedly I put more deep archive stuff on the older, slower disks and all the latest stuff on decent UltraSCSI hardware.


It makes for easy arrangements and editing - and I use the fast disks for video editing as well.


The point of this post - use old discarded kit. Rack it up in your attic (well I have to so my wife doesn't complain:) and network it all together and Bob's your uncle

I think you have to sit down and spend half the time that has been dedicated to this thread, to decide whether or not you have a "golden ear" and are capable of telling the difference between a quality MP3 and a ripped WAV. I would be willing to bet that you can't. In car, with a portable and any headphones or even on your home system (unless there's a problem with either source or some other colouring of the sound during playback due to equipment).

Personally I would find it a huge waste of time, space and money to keep WAV files when I can keep the CDs as back up. Rip times are very fast (certainly when compared to the amount of time wasted doing any of the things mentioned in this thread). The time-consuming process is the lossy encoding.

The time required to decompress the FLAC file is, more than likely, a result of the file sizes we're dealing with - try moving a file of the same size around.

Bottom line, given unlimited bandwidth and storage space, we could keep raw tracks around. Lossless compressed would still make transfers that much faster. And lossy compression, would still have benefits and possibly no audible differences. Some people forget about the "audible" part.

Bruno

That means a 100Gb worth of WAV files would take about 26 hours to compress, and a further 13hours to decompress it to WAV again. But you're right this is one way of reducing hard-drive requirements, but it doesn't deal with my particular requirements.

Are your requirements
1) Spending money on a bigass raid array.
2) Archiving your music collection in a lossless format suitable generating mp3s.

Don't let things get out of perspective. How long would it take to mp3 encode 100Gb of WAV data on an old p2 ? The 26 hours would be a one off thing as in the future you would rip straight to flac instead of to wav.

So is an extra 13 hours a significant delay compared to the time taken to mp3 encode 100Gb of wav on a p2 -333?

On a modern system you'd probably be looking at 3 seconds to decode the flac and 40 seconds for a high quality mp3 encode. And don't forget that by piping the commands togeather allot of this is going on in parallel, as much of the time taken in the flac decode process will be i/o.

If you can half you current storage requirements this way, then by the time you have doubled your music collection you can probably just go out and buy a 10Tb disc and be set for the next 10 years. Maybe buying another 10Gb in the meantime for AV stuff if you do start messing with that.

My requirements are the ability to encode mp3 to play on my eMpeg (primarily) and mp3pro for burning onto CDRW and taking to work, and low bitrate wma to use on my portable player. Since getting envolved with encoding music files, I've re-encoded several times, firstly with Xing, then with lame 3.88 , and since then lame 3.91 to take advantage of improving technologies. I don't have to, but I choose to.

My beef is not the 26 hours it would take to encode to FLAC, that as you say is a one off, I just posted that as information for the forum. It's the time to take to decode that is bothersome. I have to consider that what is an already long process to produce the end-product I want, will be even longer because of this extra step converting FLAC to WAV, and that I need the space anyway to give me the convenience (heaven forbid) of encoding all my mp3's in one hit. UNLESS there is a way to directly encode from FLAC to mp3/mp3pro/wma files, then convenience and WAV files are my requirements.

Until 2 days ago I didn't even know about the existence of FLAC. As a technology, going on what's been said, it's the equivalent of WinZip, and I'm not faulting it. It's just at present the raw ingredient used by the encoders seems to be WAV. Hopefully one day either one day mp3/mp3pro/wma encoding methods will have matured so that it doesn't change as it does, or these encoders will accept FLAC as the source material. Until then spending a couple of hundred pounds on a few disk is IMHO worth it.

...will be even longer because of this extra step converting FLAC to WAV, and that I need the space anyway to give me the convenience (heaven forbid) of encoding all my mp3's in one hit. UNLESS there is a way to directly encode from FLAC to mp3/mp3pro/wma files, then convenience and WAV files are my requirements.


as wfaulk posted earlier in the thread:

for i in `find . -name \*.flac -print`; do
flac -dc $i | lame -b128 -h - - > ${i%.flac}.mp3
done

No space wasted (all in memory) for wav data. As the flac decoding is far quicker than the mp3/whatever encoding, it'll just pile up the data in the pipe to lame while lame will start encoding as soon as it gets the first data - very little extra time spent compared to encoding directly from wav. Can be run on any level in your archive, from the whole tree to a single directory/album.

If you'd encapsulate the command in a script, it'd even look like it was a single step flac -> mp3 conversion

The above loop is unixy, but it should be possible to do something similar on other platforms (don't recall if you've mentioned what you're running)

/Michael

t's just at present the raw ingredient used by the encoders seems to be WAV

Well as mentioned somewhere, it's easy to send your flac files straight to lame without having to first generate intermediatory WAV files, just pipe the two commands togeather. The extra time for the flac decode should be insignificant compared to the time for a high quality mp3 encode.