Hello,

As you can see i'm new in the forum

After registering to the forum i read a topic comparing different lossless audio formats, as well as one predicting the future of compressed audio.
I've heard a lot of Monkey's Audio (*.ape) so i decided to make my own comparison.

I ripped KoRn - KoRn (Metal Album) to *.wav with Exact Audio Copy for digitally perfect copies of the CD. Then, i converted the tracks to *.ape with "Extra High" compression settings and to *.mp3 using LAME 3.92 with the "--r3mix" command so the output would be high quality VBR mp3s.

I compared them both in winamp 2.81 with the ape plugin. The diference between both formats imperceptible, unless, of course, in the size of the files.

So, is lossless compression really the future of audio? Lossy compression formats will always be smaller then lossless, and if the diferences between them is inaudible, who cares if my encodings are lossy or lossless? What you hear is only theorically different. In practical terms, it is (im convinced) absolutely the same, even in blind tests.

I challenge you guys to do the same. Compare both formats, using LAME encoder for mp3 with the --r3mix setting. You'll get the same conclusion i did.

I think you're a little confused.

There's no real format comparison going on here. You're saying that, to you, LAME --r3mix is indistinguishable from the original. No problem with that -- on 99% of the stuff I listen to it would probably be indistinguishable from the original for me as well.

However, there are a significant number of people for whom --r3mix is totally inadequate. The deficiencies in --r3mix were one of the motivations in the creation of LAME's --alt-presets (standard/extreme/insane), and even these cannot work around certain basic limitations inherent in the MP3 format.

They have demonstrated time and time again, using methods which produce valid statistical information (look up PCABX for a testing methodology a little more advanced than using WinAMP), that they do not lie when they say that they can hear problems.

Go here: http://www.mp3dev.org/mp3/gpsycho/quality.html
Download fatboy.wav
Encode with --r3mix
Listen to the encoding

Changed your mind yet?

Now try again with --alt-preset standard

Still interested in more of these kind of things?
Go here: http://ff123.net/training/training.html

Lossy is just that: lossy. It will never, ever be the same as a lossless file. Can a lossy file sound transparent? Of course. But it's a "doctored" version and therefore, if you want to process the file further (e.g. transcoding) you are starting off with an degraded version.

Even the king of lossy formats, MPC, is not completely transparent in every situation. That enough is a case for why lossless has some place. Lossless is perfect. There is no risk. You can reencode without fear of degradation.

--r3mix is far from perfect, as people have slowly realized. THAT is the real risk. For a time, I found 160kbps CBR MP3 to be as good as CD. However, as I trained my ears to listen better, I noticed problems and artifacts that I had originally ignored. You may find --r3mix to be "perfect" now but inadequate later on. As a result, you might find yourself reencoding your music to a superior lossy format. With lossless, you don't have to do that.

Almost everything I rip and encode gets the lossy treatment. Why? Space. With 750+ albums, I don't have the hard disk space to accomodate all of those albums in lossless format. But in several years I may be able to. Terabyte hard drives aren't that far off.

Lossless with become more common in time, but lossy will be around for a while because container formats (i.e. 700MB CD-ROM, 128MB flash card) don't swell in size year after year like hard drives do.

One description I heard on this board which struck a cord with me was, I use lossless because I'm encoding for the future.

I'm starting to do the same. I'll keep my collection of music that isn't on cd's in lossless... that way I can always encode to AAC/OGG/MPC/MP3 as though I had the original. I will keep my music in AAC+SBR form, possibly as low as 64kbs, because that sounds pretty much the same as the original to me! But... if something even better comes along in the future, which it most likely will... then I wont be penalized for having encoded my collection.. I can just go back to the "originals" and encode again!

Other than that, lossy is better as it's 2-5x smaller! But if you're looking towards the future, go lossless I think!

Right, that's exactly i wanted to say. I just used .ape instead of the original .wav to compare. If .ape is really a lossless format, then the comparing wavs or apes with mp3 leads to the same concluson



Now, maybe i am a little confused. I've heard about that --alt-presets you are talking about, but can they encode better then --r3mix? How can --r3mix be totally inadequate if the output is (in most cases [all i ever heard]) an indistinguishable compressed copy of the original file? (im talking about audible terms only, that's why i used winamp and not a file analyser).




Thats why i made the comparison using EAC to extract directly from the CD. However, you're absolutely right. Maybe if i re-encode the files to mp3 again i can hear some artifacts - i'm degrading the file.
I forgot that point



OK, the enconding is different from the original...


Please understand that im not saying that mp3 is better than lossless. I'm just saying for human hears, file formats with such a good quality/size like mp3 are "perfect". We don't need lossless to have (in audible terms) a perfect copy of almost every sound we can hear. There are some exceptions, like the examples tangent gave me, but... will lossless ever be small enough to fit our needs? I guess we will never be able to playback 5 lossless albums from a normal 650 Mb CD-R, like we do with lossy compression.

That's like saying "Is a screwdriver better than a hammer?"

Lossless and lossy compression are two different tools for two different tasks. When you need perfect archival for whatever reason, go with lossless. When you need small files (portable player, computer at work with not much hard drive space, etc.) go with lossy.

While storage and bandwidth capabilities are in general advancing at mind-boggling speeds, there will always be some situation where storage or bandwidth is at a premium. For example, just because many people have broadband internet connections nowadays doesn't mean there aren't people that are stuck with 28.8kbps modem. Therefore, lossy compression isn't going to go away anytime soon.

The point with MP3 is that the problems come in bursts. You can have a perfect sound for one minute, then a big "squeak" then no problem for 5 minute, then a whole track messed up...

To begin with, check the Recommended lame settings then you can test them with the samples at http://www.ff123.net/samples.html or the ones linked in http://www.hydrogenaudio.org/forums/index....=ST&f=15&t=3920

Double-blind testing by many users who could successfully differentiate between and original and an MP3 sample. Many of these good-listeners frequent here, and most likely are able to identify --r3mix from original WAV on killer samples.

As for lossless vs. lossy there comes a time to choose between original integrity or filesize reduction, respectively. While I'd make MP3's of my favorite CD tracks to reside on my hard disk, I would losslessly encode a master recording to be stored offline and get the slight file reduction with zero degradation.