HELP! Please URGENT (mp3 vs cd-audio comparison) Options

[SLOQshtr]

I posted this on another forum and someone sugested I should try also here:

I appologize in advance if this message doesnt fit in here!
Here is my problem: (please read it)
I have a bet going on with someone. He claims he can hear difference between mp3 and CD. I dissagree. So here's what I plan to do: He will provide a CD with 5 or 6 tracks. It will be sort of "audiophile" CD. I have to encode it to mp3 and then make 4 or 5 CDs from that, with mixed CD and mp3 tracks. Some tracks will be from CD directly and some will be decoded from mp3 to audio.  I plan to use LAME encoder, because I heard it's the best one. I don't know what exact parametrs to use, so any help would be appreciated. I suppose if I use high bitrate it shouldn't be possible to hear difference between mp3 and direcd digital copy.
My opponent in this bet has sort of "good" equipment in his house and he claims he can hear above 20000 Hz, which I doubt. Please help me, I would surely like to show him he's wrong.
Thank you in advance,
qshtr

So far my plan is to use lame 3.93 with --alt-preset insane. CD will be ripped with EAC, secure mode, low speed. I would be very grateful for any suggestion.
Perhaps I should also mention that this test will not be done on some "high-end" equipment, i suppose that price range of his equipment is 2000 -3000 EUR maximum.

Please help me win this bet.

Qshtr

[floyd]

3.93 isn't recommended. Use 3.90.2 or 3.92.

--ap insane is the highest quality preset. I think you can add -Z for even higher quality, but that might be only on --ap standard and extreme.

I wonder if this guy thinks he is going to be testing 128 kbps mp3s? Or you didn't agree on a bitrate? Sorta easy for you then

[Dibrom]

I think you can add -Z for even higher quality, but that might be only on --ap standard and extreme.

No, -Z is not what you want to use with --alt-preset insane. This preset already uses noise shaping 1, so if you use -Z it will toggle it back to the setting that you are probably not wanting to use.

[SLOQshtr]

Thank you for your answers.

We didn't specify bitrate at all, so I'm going to use highest bitrate possible. I suppose he really thinks I will do a 128kbit mp3s.

What I wonder is which are the highest frequencies encoded with ap insane. Above 20000 Hz?

I'm sure there must have been tests like this. Do you have any reference URLs for this kind of tests?

Qshtr

[twostar]

We didn't specify bitrate at all, so I'm going to use highest bitrate possible. I suppose he really thinks I will do a 128kbit mp3s.

Very sly indeed. B) If he's expecting 128kbps mp3s, I'm sure even --alt-preset standard would do the trick. But just to be safe, use --alt-preset insane.

[liekloo]

Or even the old 320 kbps setting (cbr)
The difference will be inaudible, but if you really want to be extra sure...

[Jan S.]

I already replied at the other forum:

It is very likely that he can hear the difference. It is not likely that he can hear it on any audio sample if a high quality commandline was used with LAME; though it is possible that he can hear a difference no matter what commandline was used if he finds a good killer sample.

I don't know what exact parametrs to use, so any help would be appreciated.

http://www.hydrogenaudio.org/forums/index....T&f=15&t=203&s=

e claims he can hear above 20000 Hz, which I doubt.

I doubt it too but it's not impossible.
He might be able to hear a 20000Hz signal but if he can pick if it's missing the music I can't believe.
http://www.hydrogenaudio.org/forums/index....&f=1&t=4256&hl=

[Case]

You better hope he doesn't bring you CD that is gapless, if he does no setting will help you.

[tangent]

Please help me win this bet. 

I'd say you can win the part about the 20kHz cutoff, but you cannot be sure about 320kbps MP3s...

[ancl]

Or even the old 320 kbps  setting (cbr)

Which is what --alt-preset insane is...

[SLOQshtr]

Thanx to everyone.

You better hope he doesn't bring you CD that is gapless, if he does no setting will help you.

Case: Can you be more specific? Why is that?


Qshtr

[kjoonlee]

SLOQshtr: I think he means that because MP3 doesn't have sample granularity, you might be troubled by slight pops in between MP3 files. LAME *does* have a gapless mode, but I don't really know how to use it, much less how much help it can be..

CODE

   --nogap <file1> <file2> <...>
                   gapless encoding for a set of contiguous files
   --nogapout <dir>
                   output dir for gapless encoding (must precede --nogap)

edit: LAME has + *does* == LAME *does* have (doh)

This post has been edited by kjoonlee: Dec 1 2002, 16:49

[M]

Hmm... would it be cheating to use a "freeformat" stream? A freeformat stream is a CBR MP3 which does not use the bit reservoir, but which would allow you to use a non-standard bitrate in excess of the 320kbps limitation (you can go as high as 640 kbps, if I remember correctly). The downside is that most burning software will not be able to decode such a stream on-the-fly, so you will have to decode with either MAD or LAME. Anyway, if you try "--freeformat -b640" you will have effectively doubled your bitrate, from "--alt-preset insane."

Still, this isn't highly recommended.

- M.

EDIT: The easiest way to encode gapless audio with LAME is to use Speek's frontend. That way you won't have to worry about having your "--nogap" and "--nogapout" switches in the right order. In fact, don't even bother adding them. The "No Gap" button does that for you.

This post has been edited by M: Dec 1 2002, 14:21

[ErikS]

I did this with one of my friends about two years ago. He claimed that with his expensive, super hi-fi speakers and amp, mp3 sounded like crap.

At the end of the day it turned out he couldn't even distinguish blade@128 from the original. So much for all his bashing on mp3. He never said a word about it after that.

[SLOQshtr]

Hmm... would it be cheating to use a "freeformat" stream?

No, it wouldn't be cheating if mp3 file plays in winamp. How I prepare mp3 before burning is not important for the test. All that matters is that a CD must be made from mp3s.

ErikS: I hope I will get the same result.

Keep the suggestions and opinions comming please...

Qshtr

[Jan S.]

hmm.....freeformat is highly experimental and not recommended... I wouldn't be so sure about it giving better quality than the --alt-presets.

[Case]

Thanx to everyone.

You better hope he doesn't bring you CD that is gapless, if he does no setting will help you.

Case: Can you be more specific? Why is that?


Qshtr

kjoonlee was correct, mp3 likes adding pause between tracks and this makes very audible difference. Lame's nogap feature works pretty well but it's still not perfect. It requires that mp3s are decoded with 'lame --decode' and there seems to be occasional artifacts at track changing positions.

[mithrandir]

This is silly. He claims he can hear the difference between MP3 and CD and he's absolutely right. What's the point of using API other than to realize that you are on the wrong side of the bet and that you're trying to find someway around MP3's very real limitations?

[Garf]

This is silly. He claims he can hear the difference between MP3 and CD and he's absolutely right. What's the point of using API other than to realize that you are on the wrong side of the bet and that you're trying to find someway around MP3's very real limitations?

What are you trying to say? It's not possible to get MP3's that are indistuinguishable from CD for the great majority of music you throw at it?

I think it is.

[mithrandir]

This is silly. He claims he can hear the difference between MP3 and CD and he's absolutely right. What's the point of using API other than to realize that you are on the wrong side of the bet and that you're trying to find someway around MP3's very real limitations?

What are you trying to say? It's not possible to get MP3's that are indistuinguishable from CD for the great majority of music you throw at it?

I think it is.

I think the original poster has dug himself in a hole because his bet is so generic and vague. Somebody claimed he can tell the difference between MP3 and CD. Generally? Specifically? The other bettor is going to provide some CDs as source material. We don't know what these will be. We definitely know that MP3 - like all lossy formats - audibly differ from the original source material. The extent varies from perceptual transparency to overt artifacting. We know that MP3 is particularly prone to artifacting - more than Vorbis or MPC - so if someone claims he can tell the difference between MP3 and CD, I'm not going to argue with him because there's a chance he's objectively right.

This is not about whether MP3 can be transparent on a "great majority of music". The other bettor is going to provide specific "audiophile" samples to use for this test. He's going to pick stuff where MP3 typically fails to provide transparency. And I think it's rather silly to conjure some super LAME API-modified command line to somehow make up for MP3's limitations because who in the real world uses 320kbps 2 channel lossy files anyway?

[fewtch]

And I think it's rather silly to conjure some super LAME API-modified command line to somehow make up for MP3's limitations because who in the real world uses 320kbps 2 channel lossy files anyway?

Why not? It's nearly 4.4:1 compression over the original .wav file, and if it's *really* transparent then that's certainly better than can be achieved by any lossless compression algorithm.

This post has been edited by fewtch: Dec 1 2002, 19:05

[mithrandir]

And I think it's rather silly to conjure some super LAME API-modified command line to somehow make up for MP3's limitations because who in the real world uses 320kbps 2 channel lossy files anyway?

Why not? It's nearly 4.4:1 compression over the original .wav file, and if it's *really* transparent then that's certainly better than can be achieved by any lossless compression algorithm.

It's not and it will never be. And you can transcode safely from lossless. You can't do that with MP3.

Remember, I'm not saying MP3 is crap all the time. But the first post was "other person claims MP3 != CD. I say MP3 == CD." In this context, you have to go with the other person. MP3 may equal CD in a nice chunk of cases but this does NOT seem to be what this argument is over.

[Pio2001]

Don't bother with gaps. If there are some tracks that need to be gapless, encode them in one MP3, decode them in one wav, and instert again the track markers where they were on the original.

By definition, freeformat doesn't comply with Mp3 standard. The file must come from true MP3s, so freeformat is not allowed.

[tangent]

he couldn't even distinguish blade@128 from the original

this is really sad...

[SLOQshtr]

I see we have some nice debate going on. I haven't got the time to read it all, so I'll post a little later.

I would really like to thank all who participate, especially mithrandir who has a point. I will post more details later.

Qshtr