MP3 Surround Format Debuts Today

QuadraphonicQuad

Help Support QuadraphonicQuad:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
ThomC said:
While not endorsing MP3s, I would like to say that downloadable files are 5.1's only hope.

I think title availability is 5.1's only hope. mp3 is an awful sounding format, how can anyone stand all that distortion in the cymbals? I just can't stand a 128kbps mp3. 192 is tolerable, but still not CD quality. I think the only real reason mp3 caught on like it did was because it was free. Mp3s are so commonly played on computer speakers, or through headphones on an ipod or mp3 player. I don't really think mp3 is used much on a home theater system, or much of anything that's equipped with surround sound, so honestly I really don't see mp3 surround effecting the 5.1 market that much for good or for bad.
 
ThomC said:
While not endorsing MP3s, I would like to say that downloadable files are 5.1's only hope.
Thom,
downloadable 5.1 stuff may be intresting but... when it will happen?
After all the mess Majors (and Minors) Record Companies made about the whole mp3-download story?
Mind, mp3 succeded because of the *heavily* piracy issue linked to the "sharing" concept that Majors has fought with all the ways they could. The solution is that nowadays nether dvd-a nor sacd are easily "rippable" and convertible to 5.1 whatever format, while CD was.
What 5.1 music people may share, if titles released are few, badly distribuited and possible to find only in some countries? Or us mch-enabled people are "supposed" to be the "evangelizers" of the mch downloading stuff, converting our mch music to mp3s and placing it for download on kazaa and so on? Forget it...
It's not going to happen. Will like it, but isn't going that way, and will be happy to be proved wrong in a couple of years.
 
ArmyOfQuad said:
I think title availability is 5.1's only hope. mp3 is an awful sounding format, how can anyone stand all that distortion in the cymbals?

:(

I just can't stand a 128kbps mp3. 192 is tolerable, but still not CD quality.

Variable bitrate mp3 with an *average* of 190 kbps (see --alt preset standard below) can be indistinguishable from CD quality. IME. I've also had three people (so far) attempt to distinguish the redbook from the mp3, in a blind test, using a disc I made featuring a variety of rock, jazz, and classical music; none scored statistically better than chance.

Note that even at 320 kbps constant bitrate, the most 'insane' mp3 setting available, a lot of data is being 'thrown away'.


From Hydrogenaudio.org

///
(http://www.hydrogenaudio.org/forums/index.php?showtopic=28124)

These settings require Lame 3.90 or later. Lame 3.90.3 found on this website is the recommended version. LAME 3.95.1 and later use a new VBR preset system, which is discussed in more detail [here].

Note: At a given bitrate range, the quality scale usually works to where VBR is higher quality than ABR which is higher quality than CBR (CBR < ABR < VBR in terms of quality). The exception to this is when you choose the highest possible CBR bitrate, which is 320 kbps (--alt-preset insane).


-------------------------------------------------
Recommended encoder settings:
-------------------------------------------------

--alt-preset standard (~190 kbit/s, typical 180 ... 220)

--alt-preset fast standard (~190 kbit/s, faster but potentially lower quality)

--alt-preset extreme (~250 kbit/s, typical 220 ... 270)

--alt-preset fast extreme (~250 kbit/s, faster but potentially lower quality)

--alt-preset insane (320 kbit/s CBR, highest possible quality)

For high quality on portable MP3 players, you may use --alt-preset medium (around 160 kbit/s). The medium preset is only available in the modified LAME3.90.3.

//



I think the only real reason mp3 caught on like it did was because it was free.

That, and it allowed people to store tons of music files in rather small drives.

Mp3s are so commonly played on computer speakers, or through headphones on an ipod or mp3 player. I don't really think mp3 is used much on a home theater system, or much of anything that's equipped with surround sound, so honestly I really don't see mp3 surround effecting the 5.1 market that much for good or for bad.

Video data compression (mpeg4) is used every time you watch a DVD. Data compression per se is not evil. Mpeg3 comes in a variety of flavors, some of which are suitable for critical listening, some not. The main issue for most people with stereo/HT systems, I suspect, isn't so much sound, as that they don't have any devices hooked to their home systems that can *store* and play mp3s. So they probably haven't even heard mp3 over their home systems, unless they've made a disc of them and played them in an mp3-compatible CD/DVD player. Most people don't want to run a long line from their computer to their home theaters.

If they did, they'd probably find 128 kbps CBR unacceptable, but probably would have no problem with stuff encoded as recommended above.
 
winopener said:
Thom,
downloadable 5.1 stuff may be intresting but... when it will happen?
After all the mess Majors (and Minors) Record Companies made about the whole mp3-download story?

From what they learned doing that, they should be able to add surround music files with little adjustment.

Note too that modern soundcards support surround output, so people can have computer surround setups in their work rooms as well as 'standard' ones in the HT room. Success of a surround mp3 format depends on increasing popularity of such setups, or of making a computer/HT interface simpler.

In the future, all HT receivers will probably be connectable to the Internet anyway.
They'll also probably offer storage drives of some sort, perhaps detachable for portability. Some high-end 'media servers' already behave this way. They're essentially dedicated computers grafted onto an AVR.

Alternately, and this is already happening too, there;ll be wireless data transmission between receiver and computer. (Nowadays it requires another box to act as an interface, but in the future AVRs will have that box built in) I've already tried this, and it definitely has promise. The issues for me were the poor interface software
and the lack of format flexibility (my audio files are all stored in lossless FLAC format). No wireless interface box I tried had a combination of great interface and multiformat flexibility. But that will come.



Mind, mp3 succeded because of the *heavily* piracy issue linked to the "sharing" concept that Majors has fought with all the ways they could. The solution is that nowadays nether dvd-a nor sacd are easily "rippable" and convertible to 5.1 whatever format, while CD was.
What 5.1 music people may share, if titles released are few, badly distribuited and possible to find only in some countries? Or us mch-enabled people are "supposed" to be the "evangelizers" of the mch downloading stuff, converting our mch music to mp3s and placing it for download on kazaa and so on? Forget it...
It's not going to happen. Will like it, but isn't going that way, and will be happy to be proved wrong in a couple of years.

Obviously commercial success of a surround mp3 format will also depend on companies releasing more music in surround. This can be done properly -- with new mixes -- or if they want to save money they could cheat and simply encode the output of a stereo track run through DPL II or similar DSPs.
 
Last edited:
Perceptual encoding is a psychoacoustical con trick.
It is Audio snake oil.
You cannot throw away most of the data and still expect it to be exactly the same as it was before.
This is nonsense.

Interestingly enough, the same con is now being done with pictures and video too.
If perceptual compression is really that good, then why are JPEG/MPEG so artifact riddled - and it is far easier to fool the eye than the ear.

Those who claim that 320Kbps MP3 is CD quality - well, words fail me they really do.

Last word from me on this one, as there is no way the pro MP3 mob will ever admit they got conned, so there is no point even attempting to prove different.

Imagine, if you will...
If MP3 is really that good, why are all the 2" multitrack tapes not all archived as MP3 files instead of on that high maintenance tape, or as 24/96 and even 24/192 digital transfers?
because you cannot throw away the data, that is why! It will not be the same quality.
 
neil wilkes said:
Perceptual encoding is a psychoacoustical con trick.
It is Audio snake oil.
You cannot throw away most of the data and still expect it to be exactly the same as it was before.
This is nonsense.

Interestingly enough, the same con is now being done with pictures and video too.
If perceptual compression is really that good, then why are JPEG/MPEG so artifact riddled - and it is far easier to fool the eye than the ear.

Those who claim that 320Kbps MP3 is CD quality - well, words fail me they really do.

Last word from me on this one, as there is no way the pro MP3 mob will ever admit they got conned, so there is no point even attempting to prove different.

Imagine, if you will...
If MP3 is really that good, why are all the 2" multitrack tapes not all archived as MP3 files instead of on that high maintenance tape, or as 24/96 and even 24/192 digital transfers?
because you cannot throw away the data, that is why! It will not be the same quality.

It always surprises me how deaf some people seem to be. I've heard some pretty amusing explanations on why "mp3 is better than CD". My favorite was the person who claimed since cd is 16 bit, and mp3 is 128, it's better than CD. Then there was the person that insisted that 128 is CD quality, the only reason 192 sounds any better is because it oversamples and allows for better error correction. And then there was my friend that insisted minidisc sounds so much better than CD, and claimed it was intended to replace the CD. Although, I was able to convince most people that 128kbps mp3 is less than CD quality, once I made a CD from an mp3, and A/B'd it for them, they heard it. It's just so obvious.
 
So are you guys going to pay to have my inanity meter fixed? Given that the last two posts sent it so far off the scale, it broke? The ignorance evidenced in Mr. Wilkes' post in particular is apparently so deep as to render the task of rebuttal not worth the time. I would simply recommend he do some reading up on his own, on the logic and evidence behind perceptual coding.

As for 'Army of Quad's' post, *I'm* not saying 128 kbps is CD quality.. nor is any knowledgeable person, including ANYONE WHO DEVELOPS AND TESTS MP3 CODECS. So that's simply a straw man.

But I *am* asserting that one *can* make a lossy compressed file that is audibly indistinguishable from its CD source. And I'll be happy to mail a copy of my own efforts in this direction -- a disc with a bunch of tracks containing bit-perfect rips of CD tracks, along with variable bitrate, --alt preset standard LAME encoded mp3s of the same tracks, as per hydrogenaudio.org recommendations. Each track is presented three times, first as native, then as mp3, then as 'X'. Your job is to determine whether X is native or mp3, purely from listening.

Are any of the mp3-haters up to the task? I need more data points.

FWIW, I don't archive to mp3, the reason being, it's a dead-end format -- you can't do anything else with the file except play it, without risking serious
artifacting. It';s not 'future proof'. Whereas with lossless compression,if a new , better codec is invented, you can always regenerate the original file and re-compress in the new format without fear of artifacting.
 
Ssully, I would be up for your "MP3 Challenge" if I can pick the song to be encoded. Do you have access to the Coldplay album "A Rush Of Blood To The Head"?
 
Yes, I do.

But it can't be just a single trial with one song -- you have a 50/50 chance with that. Either it has to be multiple trials with that song, or multiple trials with different songs. I can do either.

It might also be interesting to give you a set, using just one song, where the track has been encoded in either 128 kbps constant bitrate, ~190 kbps variable, and 320 constant bitrate, a well as native .wav. Or maybe three sets, one for each rate. I'll consult the hydrogenaudio folks and see whether this is reasonable.

But eitehr way let me know what you're willing to use, and we'll set it up.
 
I don't want to try and compare a bunch of different formats. I just want to take the very best that MP3 can achieve with a song I am familiar with and that I know is challenging (from an encoding standpoint) then compare it against the ripped wave file. If this one experiment convinces me I will delve further into the abyss. Thanks for doing this, BTW

Actually, if you could do the same with two of the Coldplay songs, that would be great. How about "Politik" and "The Scientist"? If we only do one, I choose "The Scientist".
 
OK. Since you specified the very best , I will encode them at --alt preset insane, 320 kbps (highest possible quality). I've never used that setting, myself, but it shouldn't be hard to implement.

Please understand that the only objectively valid way to determine whether you can hear a difference is to present the comparison in a 'blinded' fashion. Just giving you two files, and telling you which one is the .wav and which one is the mp3, and asking you if you hear a difference between them, is pointless. So I will give you the song as a standard EAC .wav rip with no encoding (A), and as an mp3 encoded .wav(B). Then 19 copies of the track (X), where each will randomly be either A or B. You report back your guesses as to the nature of X1-X19.

Since it's obvious from comparing file sizes, which file is compressed and which isn 't, I'm trusting you not to use *that* method of file identification. PM me your address and I'll prepare the CDR and send it off to you.
 
I'm cool with the experiment and you can trust that I won't cheat, but shouldn't a wave file converted from an MP3 file be the same size as the original wave file that was used as source for the MP3? The size of a 16/44.1 wave file should simply be a function of the length of the clip and nothing more. In other words, all files should be the same size so I wouldn't be able to "cheat" by simply comparing file sizes.
 
One more thing that should be part of this test, to ensure a fair one, is to NOT use any modern CD that has been brickwall limited to within an inch of it's life so that it no longer has any dynamics left.
With these titles, you would be hard put to it to notice the artifacts above the dreadful distortions produced by the mastering stage.
For this test to be a true and real one, that is actually testing what Red Book CD-Audio at 16/44.1 is capable of in terms of quality, then you must use a file that has all dynamics intact, and uses these.
A modern title that is brickwalled to the point that when you rip it and load it into a WAVE editor gives you a flat line across the top of the waveform is no test.
All you are testing then is a butchered version of the waveform that may as well already be an 8 bit file.

The actual file to be tested should be a real 16/44.1 file, not a murdered one. Or else there is simply no real complexity in the WAV, and any process can emulate modern efforts without much difficulty. It is precisely this practise that makes most people think redbook is so bad in the first place, and the only reason you can even come close on MP3 - there is nothing there to encode if there are no dynamics.

I will find a suitable file.
Or is that not acceptable?
 
Cai Campbell said:
I'm cool with the experiment and you can trust that I won't cheat, but shouldn't a wave file converted from an MP3 file be the same size as the original wave file that was used as source for the MP3? The size of a 16/44.1 wave file should simply be a function of the length of the clip and nothing more. In other words, all files should be the same size so I wouldn't be able to "cheat" by simply comparing file sizes.


You're absolutely right, and I was misremembering my last round of these tests. You can't tell which is which from looking at the file sizes when both original and mp3 are .wavs (or .cda) , for exactly the reason you cite: .wav file size is a function of length.

(However, there is still a 'cheat' available to the dedicated -- that would be to look at the files in Cool Edit or some other viewer that would reveal the reduced high frequency info in the mp3, compared to the original. ;> )
 
neil wilkes said:
One more thing that should be part of this test, to ensure a fair one, is to NOT use any modern CD that has been brickwall limited to within an inch of it's life so that it no longer has any dynamics left.


Cai wants me to use Coldplay CD tracks, which probably fit your description. However, I'd be happy to use other music tracks. I've tested people before using a variety of tracks, including acoustic jazz, vocal, and classical, in addition to pre-loudness-wars rock CDs. So far no one's been able to tell original from mp3.

The most stringent test would be to use tracks that have been identified by mp3 developers (via blind comparisons) as being especially 'tough' on encoders. This is common practice among the hydrogrenaudio.org crowd. I've been meaning to gather some of these....


With these titles, you would be hard put to it to notice the artifacts above the dreadful distortions produced by the mastering stage.


Possibly so. But 'toughness' of encoding is not correlated to quality of the original recording, AFAIK.


For this test to be a true and real one, that is actually testing what Red Book CD-Audio at 16/44.1 is capable of in terms of quality, then you must use a file that has all dynamics intact, and uses these.

Well, no, not really, since as I said, the degree og difficulty in artifact-free encoding does not seem to correlate to these classical measures of recording quality. However, I'd be happy to use whatever tracks you like, as I've said.

A modern title that is brickwalled to the point that when you rip it and load it into a WAVE editor gives you a flat line across the top of the waveform is no test.
All you are testing then is a butchered version of the waveform that may as well already be an 8 bit file.


Possibly, but bit depth is not the same as encoding bitrate.


The actual file to be tested should be a real 16/44.1 file, not a murdered one. Or else there is simply no real complexity in the WAV, and any process can emulate modern efforts without much difficulty. It is precisely this practise that makes most people think redbook is so bad in the first place, and the only reason you can even come close on MP3 - there is nothing there to encode if there are no dynamics.


This simply isn't true, IME. You can come very close -- indeed, as close as possible, which is to say, audibly identical -- when encoding a good recording of, say, a symphonic work., string quartet, etc.

(I am happy to see you allow this possibility that some CD tracks could be audibly indistinguishable from their mp3 versions, though. )

I will find a suitable file. Or is that not acceptable?

It's perfectly acceptable, as far as I'm concerned. Identifying a track that's not transparently encodable by --alt preset insane or --alt preset standard would actually be useful as a benchmark.

I think the only thing I'd want to do differently than I've done before, is limit the actual test tracks to just one or two, presenting each one enough times to draw statistical conclusions from them (e.g. 19 times). By contrast, my previous method was to use 19 different tracks -- which could conceivably swamp out positive ID of any one particularly 'tough' track in the results.
 
I chose Coldplay because of the piano-intensive music, with lots of contrasts between quiet and loud passages. Most MP3s I've heard (even presumably "good" ones) have profound "underwater" distortion surrounding the piano (especially in air and decay). Based on my limited experience (I haven't listened to a LOT of MP3s) I figure a piano is a good candidate for this experiment.

Too be honest, I don't know if the Coldplay CD is heavily compressed or not... it probably is to some extent (after all, it is a modern popular release) but I don't think it is overly so. Whatever the case, it has plenty of dynamic life left to be a good experimental candidate in this regard as well.

Finally, I'm very familiar with the music and will be sensitive to alterations in the sound. I look forward to approaching this experiment with an open mind and will be pleased to share my thoughts with the folks here.

If this initial experiment results in a positive outcome for MP3, I won't stop there. I will pursue other types and situations of music for MP3 testing, but I think this will be a good first step.
 
Go Cai.

Personally, I already know what conclusions you will come to.
And that really is all I am saying here now, for all sorts of reasons.
 
Back
Top