Well.

I came here two hours to get some more info on the quality setting in CDex, and wound reading dozens of posts here and at CDex Forums, in which time I learned a whole hell of a lot...

I still have a few questions, though. I know some have been answered (at least to some degree), so please don't get mad at me for being curious or incredulous.

Anyway.

I've been living a long time with 128CBR/VBR for my personal collection. Never really had problems with sound in general, at least as long as I was listening to my stuff on cheap PC speakers. With storage becoming so cheap, I decided to re-rip from scratch, also wanting to archive my entire CD collection in the process. (Not sure if I want to sell my CDs post-ripping, but that sounds like an intersting idea.) I started using 192CBR, and did some listening tests on a decent Hi-Fi setup and decided I wasn't completely happy. Plus I thought since I was archiving, it would be best to go as high as possible, so I thought about going all the way to 320.

Then I came here, and all my preconceptions were blown out of the water. Here's my questions:

1) I admit I was brainwashed by the anti-Joint Stereo crowd. Having said that, one of the more cogent explanations (if I remember correctly) was that it messed up the stereo separation of the vocals or somesuch. Any credence to that? What exactly is Joint Stereo anyway?
2) Is VBR really that much better than CBR? Does it make a damn bit of difference if I'm going up to 320? For that matter...
3) Is 320 too high? Am I just wasting my time? I'm not worried about space as CDRs and HD space is cheap , and getting cheaper.
4) I've read quite a few posts that seem to prefer EAC to CDex. What difference does it make if they're using the same encoder?
5) Why would I want to add a CRC to my Mp3?

Thanks in advance for all the replies!

Captain Splendid

Hi there

1) joint stereo looks for similarities between the 2 stereo channels (which are very common). The encoder uses that info to reduce the bits used for stereo information and spends those saved bits on more useful things then. When it is done right, you cant hear a difference to "real" stereo and the extra bits improve overall quality. When it's done wrong, the stereo image is damaged. People on this board who know a lot more than me state that joint stereo is safe when using lame --alt-presets, and it's default in all -alt-presets also.

2) a constant bitrate of 320kbit/sec (--alt-preset insane) is the maximum possible quality for mp3. Variable bitrates (--alt-preset standard/extreme) might give the same audible quality, while the average bitrate is lower. Check out the sticky "recommended lame settings" here for more info

3) constant bitrate encoding is faster than VBR, for high CBR bitrates like 320k even a lot faster. If you want ultra-high speed use "lame -b 320 -h -m s", for "normal" high-speed use "lame --alt-preset insane" which should give slightly higher quality in theory but needs more time for joint stereo and other tweaks.

4) depends on how scratched your CD is. for a clean "like-new" CD, both are perfect in their secure mode, for badly scratched CDs try both and use the result that is less broken

5) CRC are totally useless in mp3 and waste bits you could use for encoding quality.

Hope that helps a bit
what I read from your text is that you might want to go for:

-"lame --alt-preset extreme": the best mp3 VBR quality, perfect for archives and PC playback, possible problems with hardware players though

-"lame --alt-preset insane": the 320k CBR overkill no problems except wasting some storage space

-if you have an eye on ecoding speed and dont wanna waste too much space and maybe wanna play the stuff on a hardware player later, you can check out "lame -b 256 -h -m s" (256k CBR, high quality, no joint stereo) also.

Very concise answer! That helped a lot, and basically tells me to go for 320. Neither space or encoding time is an issue for me now (got two 120GB HDs on order for expansion of my mp3 server), and as time goes by, will be even less. Believe it or not, I had actually considered going for full .wav files for a while.

One more question:

1) I'm a little unfamiliar with all this command line stuff. Basically I'm looking to use 320CBR Stereo, with quality=o, right?

Thanks again!

Captain Splendid

lame.exe --alt-preset insane

is the commandline for your 320k CBR, dont use other switches, it's all-inclusive. best used with lame 3.90.2, the link to it is in the "recommended lame compiles" sticky topic. Have fun

If you want to encode directly out of EAC when ripping (good idea), go to "compression->external compression" preferences tab, select "user defined encoder", point the path to lame.exe, and enter "--alt-preset insane %s %d" (without the " ") in the command line options.
Then select all tracks you want to read and press the "mp3" button and you're done

Very good answer from sony666.

Joint stereo really improve quality even at 320.

As told sony666, use lame 3.90.2 --alt-preset insane. I use it and I never complain (I'm really maniac about sound quality).

Peace

That's what I'm trying to figure out - even at 320, JS is still necessary? I would have thought that at that bitrate, the compression wouldn't need any any more "help".

(Hoping I don't sound too stupid on that one)

Captain Splendid

you may also want to consider MPC... or if you have the space, one of the lossless codecs before you sell off your originals...

mpc can produce equal and higher quality files at lower bitrates (ie. smaller file sizes)...

the lossless codecs are obviously perfect copies... but you'll be seeing anywhere from 200-400mb on average per cd, as oposed to 100-150 with 320kbps mp3, or a little less with mpc braindead... (which is in itself, overkill)

just figured i'd throw those options out there if you weren't already aware...

all of them are pretty easy to use & quick compression...


if you do stick with mp3s... be sure to use one of the presets... you can't go wrong with anything from standard up... (personally i use insane if i have to encode a mp3... but i'm just nuts like that )

Don't be affraid to use JS, it's better then simple stereo. There is no catch in JS, it does not alter the stereo image, it's simply a more efficent way to save stereo information.

May I ask, what %s and %d mean?

A search yielded this thread but I understood nothing from it

Hehe Well, that's about the LAME command-line output, not EAC...

%s and %d are EAC parameters. From the FAQ.TXT in the EAC documentation folder:

-q0 doesn't actually give the best quality in most LAME versions. Use --alt-preset insane, it picks the right values.

%s and %d are used in EAC and quite a few other ripping programs too. Encoders work through commandlines, like DOS programs. In DOS, you can pass values from one batch or program to another, using a umber or letter to identify the value to be passed. The more or less common standard in ripping programs is to use %s and %d to denote the source file address and the destination file address for the file to be processed.

Actually, it's quite surprising that MPC could actually sound better than any other codec at 128 kbps.

I always thought that MPC is the worst at bitrate 128 kbps, but I am wrong, it sounds superior to other codecs at 128 kbps!

you might consider using MPC or AAC instead of 320 kbps MP3, both MPC and AAC is safer than mp3.



I think you should consider using Monkey's audio or OptimFROG , then you don't have to use lossy anymore. You could save 10-20 megs for other songs that you intend to compress it to lossy

Yes, even at 320, it helps.

Don't make the same mistake I made in ripping my first mp3 in 320 stereo. When I hear a stereo mp3, in can hear two differences even at 320:
· High frequencies are weak
· When a voice is right in the middle on the original song, in stereo mp3 the voice is somewhere in the middle (really boring when you use a headphone), in JS the voice is still perfectly in the middle.

When it was possible, I re-ripped my bad stereo mp3 with --alt-preset insane. For the rest I’m stuck with the old mp3 (stereo) .

So why bother you, it doesn't cost anything and it audibly increases the quality...

Cheers

Yes, for the reason sony666 mentioned in his first reply. In Simple Stereo mode, each channel gets encoded "separately", which leaves 160kbps per channel - we all know that 160kbps MP3 isn't nearly enough to encode all samples transparently. Joint Stereo can save bits by throwing out redundant information contained on both stereo channels, thus increasing quality.


About CDex vs. EAC: I would strongly recommend not using CDex if you're concerned with quality. Its so-called secure/paranoia mode doesn't catch all possible read errors and thus often leaves you with a bad rip, even though it was reported as "OK". And its Paranoia mode is totally useless if you have a drive that caches audio data - it will always report all rips OK, even if you're ripping a CD that you've been skating on.

[As you probably know, the Paranoia mode works by re-reading each sector at least twice and comparing the results - if those don't match, then it is assumed that a read error occured and the passage is re-read until a constant output is achieved. The thing is, if your drive caches the data, the re-reads will *always* produce identical results, because the data is coming out of the cache!]

EAC can handle drives that cache by clearing the cache before attempting to re-read any sectors.


One more thing: If you're going to sell your CDs, you won't want to leave yourself with just an MP3 copy - especially if you've got live albums or other albums whose tracks are linked together without gaps. MP3 introduces gaps of silence at the beginning and at the end of files - a problem that you can eliminate if you use Winamp (with the MP3Splice plugin), but if you're planning to use the IMHO superior foobar2000, you're going to have to live with ugly gaps between your tracks.

I'd consider using MPC if I were you (probably at "--quality 6") - it provides better quality than 320kbps MP3, is a lot smaller (~200kbps), provides truely gapless encoding/playback, and will transcode to MP3 files you might need for a hardware player perfectly.

CU

Dominic

The only valid reason to use a lossy codec, such as mp3, is to save space.

If you have plenty of free space, you should consider the full wave equivalent - a lossless codec, such as FLAC, WavPack or Monkey's Audio.

I will not post a list of advantages/disadvantages. If you have already made up your mind, there is no need to. If not, you should be able to figure them out yourselves - at least those relevant to you.