Xing new ? Options

[user]

Hi all,

a friend has some songs encoded by Xing new, quality 256 CBR, all frames mj, mid side !

What do you know or think about that quality ?

I am going to show him advantages of Lame.

[westgroveg]

I think everyone knows that Xing is one of the worst quality mp3 codecs around as proven at WWW.R3MIX.NET by Roel, I havent seen any tests on a new Xing codec as of yet

[JohnV]

Actually using sine sweeps and graphs like it's done at r3mix.net doesn't prove much about psychoacoustic codecs.

Encoders are tweaked and properly tested using listening tests, not using sine sweep graphs... However, there's no question that Xing is not very good, based on listening tests.

[deranger]

Hi, All !

Not long ago I've downloaded some rare tracks which some guy was kind enough to share in Audiogalaxy, but stupid enough to encode with xing 256kbps. Well, even with this bitrate I can hear artefacts in xing encodings

I really want to burn those tracks to cdr, 'cause I think I'll never find copies of them of better quality ! So, did anyone practised "restoring" or "improving" quality of decoded xing encodings ? Can U recommend some common things I can try to do (lowpass, eq, etc.) ???

[Dibrom]

Originally posted by deranger
Hi, All !

Not long ago I've downloaded some rare tracks which some guy was kind enough to share in Audiogalaxy, but stupid enough to encode with xing 256kbps. Well, even with this bitrate I can hear artefacts in xing encodings

I really want to burn those tracks to cdr, 'cause I think I'll never find copies of them of better quality !  So, did anyone practised "restoring" or "improving" quality of decoded xing encodings ?  Can U recommend some common things I can try to do (lowpass, eq, etc.) ???

Once the information discarded by the psychoacoustic model of a lossy encoder is gone... it is gone for good, you can't just get it back with some sort of processing. Unlike things which you can do to restore analog sources such as tape or vinyl, you can't do the same thing with mp3. Looks like you are out of luck if you have no other sources for the tracks you are looking for..

[Volcano]

Actually using sine sweeps and graphs like it's done at r3mix.net doesn't prove much about psychoacoustic codecs. 

Encoders are tweaked and properly tested using listening tests, not using sine sweep graphs...

Oh come on, John, *please* stop running down Roel and his site all the time. Since this signature thing happened and since this forum has been up, you have been doing that quite frequently.

This is from r3mix.net's "Critique" section:

13. For the frequency sweeps, you're doing them all wrong!

For the frequency sweeps, you're doing them all wrong!  You should use a logarithmic sweep, 20 seconds or longer.  The problem with a 10 second linear sweep is that it zips through the ear's three most sensitive octaves in less than 1 second. A log sweep gives each octave equal time.  Use logarithmic scale for displaying spectrum as it gives each octave equal length (on the graph's horizonal axis).  Otherwise you will be guilty of grossly misrepresenting the data and contributing to the growing body of misinformation and disinformation on audio encoding.

Thank you for your honest and direct comments. You are obviously more educated about these issues then myself, but not once I use the sweeps to measure quality of an encoder. I only use them to show obvious flaws in, for example, the Xing encoder.  I am aware that I could have taken the logarithmic rather than the linear basis to display to represent a more accurate model of hearing distribution, but this would still not make that sweep comparison a valid encoder-benchmark. With my method, all that are distorted sound and look distorted, and all that sound ok, look ok. I don't want to make a pseudo-valid scientific test, because strictly seen a more deviant graph does not necessarely mean a worse encoded piece.  With mp3 you never know what frequencies are masked by others, so a sudden drop in a curve might not say all that much.  Also take in account that mp3 encoding itself divides the frequency spectrum in equal shares rather than logarithmic.

The graphs are not to be measured, they're to be interpreted with plain and simple logic.

Sorry for coming off the topic, but I just think that remark wasn't right. And let's not start this Project Mayhem vs. r3mix discussion again here, that's certainly not my intention.

CU

Dominic

[westgroveg]

Originally posted by user
Hi all,

a friend has some songs encoded by Xing new, quality 256 CBR, all frames mj, mid side !

What do you know or think about that quality ?

I am going to show him advantages of Lame.

If your looking for best quality lossy codec I would go for MPC with --insane present
and if you need to use the Mp3 format I would use Lame with --dm insane (or xtreme). Both presents are tuned for highest quality and give a much higher quality audio file (and smaller) than Xing 320kbps CBR with stereo;)

[westgroveg]

Originally posted by deranger
Hi, All !
I really want to burn those tracks to cdr, 'cause I think I'll never find copies of them of better quality !  So, did anyone practised "restoring" or "improving" quality of decoded xing encodings ?  Can U recommend some common things I can try to do (lowpass, eq, etc.) ???

Use Lame, Fhg decoder not Xing.

[JohnV]

Originally posted by Volcano
Oh come on, John, *please* stop running down Roel and his site all the time.

Umm, what exactly wasn't right in my remark? You don't have to believe me, ask any developer.
Ask how much sine sweep are used vs listening tests. Ask how much value they would give to sine sweep test when testing psychoacoustics and true audible quality.

I'm sure you'll belive Robert, Monty,Ivan or Andree, if not me.
Well I already know their opinions because this is widely discussed issue, but go ahead and ask.

[Volcano]

Hey don't get upset

My point was that the sine sweeps are never used on r3mix.net to prove the actual sound quality of the encoder (which is what you seemed to assume) - they are only used to show one obvious flaw in Xing (which is exactly what the article from the Critique section says).

CU

Dominic

[deranger]

Hi, All !

Not long ago I've downloaded some rare tracks which some guy was kind enough to share in Audiogalaxy, but stupid enough to encode with xing 256kbps. Well, even with this bitrate I can hear artefacts in xing encodings 

Once the information discarded by the psychoacoustic model of a lossy encoder is gone... it is gone for good, you can't just get it back with some sort of processing. Unlike things which you can do to restore analog sources such as tape or vinyl

Btw, what do you think of various "mp3 playback improvement" plugins, such as DFX ! I've never tried one, so I'd like to hear your opinions ...

P.S. I didn't get how to insert that "Originally posted by ..." string ? Is there an automatic way to do so ?

[westgroveg]

Originally posted by deranger
Hi, All !
Btw, what do you think of various "mp3 playback improvement" plugins, such as DFX !  I've never tried one, so I'd like to hear your opinions ...

Ive had DFX for a while, for some music I like it (although it can make sound more artifical in others) but in general i think its for people who have poor sound cards and want more because it gives the illusion of more dynamic sound.
P.S I would not use sound enhancement for archiving, just use them for playback.

this is also a good plugin:

http://www.winamp.com/components/download_...mponentid=38489