Why would anyone use a format other than mp3? Options

[Jibaholic]

If --alt-preset standard is transparent to 99.9% (or whatever) of users, then the only benefit for the other formats is size and encoding speed.

As for size, A 100G hard drive costs about $200 bucks. That means a 100meg rip requires 20 cents of disk space. The CD itself costs $10 or $15. So forget about the other formats being smaller at transparency. It's a non-issue.

So the only advantage I can think of for other formats is encoding speed. Does that really justify their existence? Or am I missing something?

P.S. This is my first post so it could be a really naive question. But I've read all the codec holy war threads and haven't seen anyone address this point.

edit: sorry new user -- I've done a lot of searches and read the stickies but I hadn't looked at the main traffic. There is already a codec comparing thread so if I mod wants to move this into that thread, please go ahead.

This post has been edited by Jibaholic: Jul 11 2003, 18:42

[sthayashi]

In a few words: Streaming. Sharing and portability

One thing that AAC, Ogg Vorbis, and WMV are doing is improving the lower end bitrates.

-aps mp3s tend to be very large. Even with the -Y switch, they clock in at around 160kbps. At around 112kbps, mp3s sound horrible. 64 is just about out of the question.
However, the other codecs I mentioned above are focusing in on that range specifically. They sound MUCH better than mp3 there, and thus are better suited for streaming online, or even just transferring.

Hope that clears things up

[Jan S.]

MP3 can't be gapless.
No standard tagging system (id3 is not in the specs).
Other formats have way better VBR so less bits are wasted where they are not needed.
Slow decoding.
Limited to 320kbps.

[evereux]

Other lossy formats provide "proper" gapless playback too.

Edit: Sorry Jan S just beat me to it.

This post has been edited by evereux: Jul 11 2003, 18:54

[Peter]

No sample-accurate seeking.

[hödyr]

No perfect Replaygain support (when not using FB2k + APE Tags).

[Volcano]

But I've read all the codec holy war threads and haven't seen anyone address this point.

Actually, it has been adressed about a million times.

The only argument for still using MP3 is the amount of hardware and software support it has got - unfortunately, that's a heavy argument. In basically all other aspects, other formats are superior (gapless playback, quality per bitrate, tagging, ReplayGain support, ease of encoding etc.). I suggest you search the forum for more details on each point, because, believe me, this subject really has been beaten to death here.

[EDIT] Oops. That's 5 replies in the time I was composing this message. [/EDIT]

This post has been edited by Volcano: Jul 11 2003, 18:56

[Covak]

If mpc gives me better quality in smaller files, why not?

Not to mention the better tagging system, it's gapless, ReplayGain support...

[atici]

But I've read all the codec holy war threads and haven't seen anyone address this point.

You read them all? Then you probably know what you'll hear:

• MP3 is not gapless.
• MP3 is not efficient in encoding high frequencies.
• MP3 does not have nice replaygain support (AFAIK).
• MP3 is an almost obsolete codec considering the latest developments in the world of psychoacoustic codecs.

And on top of these you're missing a major point IMHO: what is transparency? i.e. non-distinguishability from the original by ABX test, right? But how good of a measure is that to verify how much loss occurs? In 99.9% of the tracks, 99.9% of the people might not be able to hear a difference. But what if on one single day with one track you could hear it?

And here comes the counter argument: If I won't use any other device to play my files than my PC (which I have a high-end equipment attached), why would I ever touch mp3? I know there're other lossy codecs (mpc for instance) that would make use of my bits much more efficiently and would preserve the high frequencies better.

In the end lossy codecs is a world of compromises and preferences. It all depends what is important for you and what compromises you want to make. It is all about how perfectionist you are. People on this forum usually want to make sure they lose as little amount of quality as possible.

Otherwise your argument can be furthered like this:
Oh yeah well why can't we as well use a nonsecure ripper, we won't be able to hear the difference in 99.9% of the cases. Well why do we need expensive speakers, I can hear my music goddamn well on these $10 Creative crap. Why do we ever need to listen music? What does it add to my life? I can as well go out and have fun.

It is about how much pleasure you want to make out of your music. If all these meticuluous procedures and expensive equipment adds pleasure to your life then why do you ask such a question? If not too bad then why are you in this forum?

This post has been edited by atici: Jul 11 2003, 19:17

[Jibaholic]

[*]MP3 is not gapless.

[*]MP3 is not efficient in encoding high frequencies.

[*]MP3 does not have nice replaygain support (AFAIK).

[*]MP3 is an almost obsolete codec considering the latest developments in the world of psychoacoustic codecs.

gapless is a good point.

* mp3 can't be that innefficient at encoding high frequencies or it wouldn't do be transparent. I suppose that is where the extra space is taken up, but I think I've done a good job showing why that is a non-issue
*The mp3gain website says its based on replaygain.

And on top of these you're missing a major point IMHO: what is transparency? i.e. non-distinguishability from the original by ABX test, right? But how good of a measure is that to verify how much loss occurs? In 99.9% of the tracks, 99.9% of the people might not be able to hear a difference. But what if one one track you could hear it?

Well that is what I was driving at with my post -- the only real selling point of the other formats is to eek out that last one tenth of one percent.

It is about how much pleasure you want to make out of your music. If all these meticuluous procedures and expensive equipment adds pleasure to your life then why do you ask such a question? If not too bad then why are you in this forum?

I had planned to use ogg before learning that mp3s encoded with LAME could be transparant.

This post has been edited by Jibaholic: Jul 11 2003, 19:22

[Peter]

MP3 has one more problem: availability of free decoding libraries. Most of freeware developers need to choose between MAD (supposedly bug-free and clean, but under GPL, so it's not really reusable in non-GPL projects), and mpglib (LGPL but latest official distribution is a buggy piece of shit).
Vorbis libraries have open licensing and are easy to use. Same goes for MPC (there's no standardized MPC library but decoding source from Winamp plugins is bug-free and quite easy to reuse).

[Jan S.]

*The mp3gain website says its based on replaygain.

Again. there's no standard way to do it with mp3 files. Old solution was changing the file itself which was not the best solution. now ppl use ape2 tags...but only a few programs support this.

[Jibaholic]

So there are definitelya lot of warts even if you can ultimately get transparant audio. That makes sense. Most of the codec wars threads seem to be about which has the best quality.

I do listen to shoutcast fairly regularly, a streaming ogg or mpc radio station could be really cool. Now I'm wondering - is there already such a thing?

[atici]

* mp3 can't be that innefficient at encoding high frequencies or it wouldn't do be transparent. I suppose that is where the extra space is taken up, but I think I've done a good job showing why that is a non-issue

No, that's not true. Even Dibrom, one of the developers of LAME mentioned many times about inherent setbacks of MP3 as a codec to encode high frequencies.
To be able to hear high frequencies and to be able to distinguish it in real music is the point. Even though high frequency loss is inevitable with MP3, what you mean by transparent is you won't be able to hear the difference for that particular sample. Even though many people on this forum can hear pure 18kHz tone it's not distinguishable occuring in a piece of real music.

I'd suggest you encode some music into MP3 and calculate the difference from the original as a separate wave file (there're tools that could do that like CoolEdit I guess). Play the difference file. When you listen to the difference, would you consider this loss in your music as tolerable?

And there're also hard to encode instruments, like harpsichord which I listen a lot. I don't want to risk any major quality loss to the harmonics. That's why I avoid mp3 like death. You have done nothing as to showing anything because you cannot. Not just because you apparently didn't read the discussions carefully as you claimed you did, but also there're real issues with MP3 that cannot be addressed and is inherent with the construction of the codec.

*The mp3gain website says its based on replaygain.

That's not a native support. It is basically a hack, your player wouldn't most probably support it.

Well that is what I was driving at with my post -- the only real selling point of the other formats is to eek out that last one tenth of one percent.

Well, the other formats have a lot of selling points. What does MP3 have? Nothing for me. Even if you use 320kbps MP3 files (which is twice as large as MPC standard), the trouble with encoding high frequencies will be there. Do a test and see for yourself as I said. Make your difference (Original)-(Encoded) as a sample file and listen to the difference only. Then you'll understand what exactly is lost.

This post has been edited by atici: Jul 11 2003, 19:50

[Jibaholic]

I'd suggest you encode some music into MP3 and calculate the difference from the original as a separate wave file (there're tools that could do that like CoolEdit I guess). Play the difference file. When you listen to the difference, would you consider this loss in your music as tolerable?

If I can't ABX the difference, I would have to say yes.

[atici]

If I can't ABX the difference, I would have to say yes.

ABXing the difference and whether there's an audible difference are different issues as I said. The part you're quoting from me stated that point. If you make a pure difference file as (original)-(encoded) you can hear the difference only and it'll be quite audible. Now when you do that test before you go and puke, reconsider whether that loss is tolerable.

If you're happy with what you do then use MP3, go home and be happy. It's you asking the question "Why do people use other codecs?" and we explain to you.

As I told you before, the issue is: how much listening to quality music adds up to your life. If you don't care or you think time and other kinds of investment towards achieving higher quality is not worth it and "Ignorance is bliss" is your motto: then do what you like.

This post has been edited by atici: Jul 11 2003, 21:31

[Daybreak]

I don't think flaming him is really the way to go, I mean, most users tend to ask this sort of questions at certain points of time...

Anyway, to the original topic ... To summarise briefly, if you have or plan to get a portable player ( memory/disk/optical) anytime soon, MP3 is pretty the only choice you have.

If you play your music exclusively on the computer, you would be much better served with any of the other alternative 3 lossy codecs, or even the lossless ones.

As for the shortcomings of MP3, IMHO, the only 2 that REALLY matter are (a) lack of gapless and (B) the confusion over a proper tagging standard ( ID3v1.x, ID3v2.x, APE2... lack of official Replaygain tags ). To me, quality/bitrate and age of codec are pretty much moot points, especially if you're NOT trained to hear artifacts.

[saratoga]

But I've read all the codec holy war threads and haven't seen anyone address this point.

You read them all? Then you probably know what you'll hear:

• MP3 is not gapless.
  
• MP3 is not efficient in encoding high frequencies.
  
• MP3 does not have nice replaygain support (AFAIK).
  
• MP3 is an almost obsolete codec considering the latest developments in the world of psychoacoustic codecs.

And on top of these you're missing a major point IMHO: what is transparency? i.e. non-distinguishability from the original by ABX test, right? But how good of a measure is that to verify how much loss occurs? In 99.9% of the tracks, 99.9% of the people might not be able to hear a difference. But what if on one single day with one track you could hear it?

And here comes the counter argument: If I won't use any other device to play my files than my PC (which I have a high-end equipment attached), why would I ever touch mp3? I know there're other lossy codecs (mpc for instance) that would make use of my bits much more efficiently and would preserve the high frequencies better.

In the end lossy codecs is a world of compromises and preferences. It all depends what is important for you and what compromises you want to make. It is all about how perfectionist you are. People on this forum usually want to make sure they lose as little amount of quality as possible.

Otherwise your argument can be furthered like this:
Oh yeah well why can't we as well use a nonsecure ripper, we won't be able to hear the difference in 99.9% of the cases. Well why do we need expensive speakers, I can hear my music goddamn well on these $10 Creative crap. Why do we ever need to listen music? What does it add to my life? I can as well go out and have fun.

It is about how much pleasure you want to make out of your music. If all these meticuluous procedures and expensive equipment adds pleasure to your life then why do you ask such a question? If not too bad then why are you in this forum?

I'll play devil's advocate:

[*]MP3 is not gapless.

This doesn't bother the majority of listeners, and for those that care theres always cue files. I realize these won't work on a new Ipod, but neither does vorbis or musepack

[*]MP3 is not efficient in encoding high frequencies.

If its transparent, its transparent.

[*]MP3 does not have nice replaygain support (AFAIK).

Works fine in Foobar2000 for me. Then theres always MP3gain or whatever.

[*]MP3 is an almost obsolete codec considering the latest developments in the world of psychoacoustic codecs.

This isn't actually a fact, its just your view. Anyway in light of the fact that its generally transparent, what does this matter even if it were true?

In 99.9% of the tracks, 99.9% of the people might not be able to hear a difference. But what if on one single day with one track you could hear it?

You should be using lossless if you require >99.9% transparency.

Note: The instant MPC gets good portable support I'm done with MP3 forever. I'm just curious what people have against MP3.

[saratoga]

Well, the other formats have a lot of selling points. What does MP3 have? Nothing for me. Even if you use 320kbps MP3 files (which is twice as large as MPC standard), the trouble with encoding high frequencies will be there. Do a test and see for yourself as I said. Make your difference (Original)-(Encoded) as a sample file and listen to the difference only. Then you'll understand what exactly is lost.

All this tells us is that MP3 is a lossy codec. If you want lossless, you should use a lossless codec.

ABXing the difference and whether there's an audible difference are different issues as I said. The part you're quoting from me stated that point. If you make a pure difference file as (original)-(encoded) you can hear the difference only and it'll be quite audible. Now when you do that test before you go and puke, reconsider whether that loss is tolerable.

No they're not. If you cannot ABX anything, its because there is no audiable difference to hear.

Edit: grammer.

This post has been edited by Mike Giacomelli: Jul 11 2003, 21:07

[nyarlathotep]

The basic rule I use concerning transparency: just believe what your ears can hear. Transparency is very subjective and we could spend days and days talking about it endlessly.

Anyway, if you do use a portable player, there are not many codecs to choose from.

[Jebus]

I'd suggest you encode some music into MP3 and calculate the difference from the original as a separate wave file (there're tools that could do that like CoolEdit I guess). Play the difference file. When you listen to the difference, would you consider this loss in your music as tolerable?

Come on man, that isn't fair. This is a similar argument to comparing spectral graphs the r3mix way. The goal of a perceptual lossy codec is to discard the stuff you won't miss.

For instance, if you use -Y, most people won't notice the high frequency sounds missing. but if you diff'd this with a full sfb21 --alt-preset standard you would certainly hear the differences in the diff file, because they aren't being masked out anymore.

Oh, and mp3gain IS a hack, but it is also compatible with every single mp3 player out there. Don't mislead the newbies!

I just use MP3 myself because I primarily listen on an HD portable, and there really ISN'T a great big reason go through the headache of using two different formats when my MP3s are, to be honest, just fine. So in this sense I agree with the parent post.

This post has been edited by Jebus: Jul 11 2003, 21:21

[atici]

No they're not. If you cannot ABX anything, its because there is no audiable difference to hear.

Not true. That's why if you reencode multiple times the loss becomes more apparent. The difference between encoded lossy and original is not distinguishable only when superimposed on the resulting music. Otherwise the difference is perfectly audible by itself, especially when the original is passed through a lowpass filter...

Oh, and mp3gain IS a hack, but it is also compatible with every single mp3 player out there. Don't mislead the newbies!

Really? even the hardware players? I don't think it is.

This post has been edited by atici: Jul 11 2003, 21:24

[smok3]

also some encoders are quite faster than Lame. (at least mpc and vorbis.)

[ErikS]

Really? even the hardware players? I don't think it is.

I should be. Do you have an example of one which doesn't play an mp3gained file?

But to try and stay on topic:

If you have two options where one is better than the other in every single aspect that is important to you such as quality, features (tagging, replaygain, gapless playback), speed than the other plus it's cheaper (because it takes less harddrive space), then what do you choose? The choice was simple for me...

This post has been edited by ErikS: Jul 11 2003, 21:52

[Peter]

mp3gain does work with all players, because it modifies mp3 data itself, so it doesn't need any kind of support on player side.
Now please stop trolling (your posts about listening to "difference files" are BS too) and putting claims about things you have no clue about.