HydrogenAudio

Lossy Audio Compression => MP3 => MP3 - General => Topic started by: David Nordin on 2001-09-30 11:07:01

Title: What should become of LAME?
Post by: David Nordin on 2001-09-30 11:07:01
*removed*
Title: What should become of LAME?
Post by: Jon Ingram on 2001-09-30 11:44:09
Quote
Should we convince Robert and the fellas to go for Vorbis? or MPC 


It's very hard to work on enhancing a coder for a format if you don't even have the specifications for the format... so MPC is out.
Title: What should become of LAME?
Post by: Dibrom on 2001-09-30 12:03:09
Quote
Originally posted by Jon Ingram

It's very hard to work on enhancing a coder for a format if you don't even have the specifications for the format... so MPC is out.


Actually very soon now there will be more programmers joining in with the MPC development team, and I believe some of them are well known within the audio community, even the LAME community.  So even though the encoder may not be open source that doesn't mean that others can't help with the development of this format.
Title: What should become of LAME?
Post by: Garf on 2001-09-30 12:11:07
I would hope most go work on Vorbis. I'd prefer to have a totally free format even it if takes just a bit longer before it reaches MPC quality, than a patented format where future versions won't be free (or will they? last I heard was shareware)

Also, the openness of Vorbis makes for more tinkering possibilities. I need to write a 'Vorbis tuning for beginners'  someday

Hmm, I assume those working on MPC get a share of the shareware income? That might be a good incentive I guess...

What improvements will be made in the next few MPC versions? I thought the MPC developer said that technically the format was nearing the end of its development, so I wonder what they are going to improve.

--
GCP
Title: What should become of LAME?
Post by: stoff on 2001-09-30 14:27:04
Interesting indeed...

I hope - like Garf - that most of the work will go into Ogg Vorbis.

If Vorbis really has the possibility to reach MPC quality I don't see why anyone should prefer MPC in the long run. Competition is usually a good thing but I don't see how it'll be a good thing to spread the the effort to much since MPC and Vorbis wont stand a chance to succed at the same time IMHO.

Regards stoff
Title: What should become of LAME?
Post by: elfin on 2001-09-30 14:38:20
If LAME team and Vorbis team join forces. I think it's a next logical step.
Title: What should become of LAME?
Post by: pho3b3 on 2001-09-30 14:42:21
Quote
Originally posted by Garf
Hmm, I assume those working on MPC get a share of the shareware income? That might be a good incentive I guess...
GCP


If such guys as Buschmann, Shibata, Bouvigne and too many others could work together that would be great! 

Thus shareware would be the very reward for their herculean work.
Title: What should become of LAME?
Post by: pho3b3 on 2001-09-30 14:50:20
why not concentrate efforts in a really promising technology?

why not joining forces and make some patent holders' rules to be overcome by popular wish?
Title: What should become of LAME?
Post by: Garf on 2001-09-30 14:53:20
Quote
Originally posted by pho3b3


If such guys as Buschmann, Shibata, Bouvigne and too many others could work together that would be great!  

Thus shareware would be the very reward for their herculean work.


Well, LAME was free, Ogg is even free-er.

The reward for their work has always been the appreciation of the users.

If that's not enough for the MPC guys, so be it, but I[/b] am not going to pay to use MPC. I will pick a format that is really Free.

--
GCP
Title: What should become of LAME?
Post by: CiTay on 2001-09-30 15:15:51
You talk about LAME like it's already dead... :crying:

We're just about to see a new LAME stable version soon, and you guys make plans what to do with the developers once they abandon this great project? :enraged:

There is still something going on at LAME, and just because it doesn't make big steps every other week doesn't mean that it's over. :rant:
Title: What should become of LAME?
Post by: Garf on 2001-09-30 15:27:58
Quote
Originally posted by CiTay
You talk about LAME like it's already dead... :crying: 


What? Didn't they tell you?

MP3 IST DEATH! WMA IST GOD!

:freaked:

Quote
We're just about to see a new LAME stable version soon, and you guys make plans what to do with the developers once they abandon this great project? :enraged:


Well, it's not us who get to decide of course, but several of the LAME core have been talking about leaving the project.

That doesn't mean it dies. In fact, its already reached a high enough level of maturity it probably never will. But it does seem to have gotten over its top. That doesn't take away from the fact that, as a very mature project, it's certainly very powerfull.

Fixing gapless and the LAME tag are the two last outstanding issues I  know of. Perhaps it would have been nice if nspsytune could be improved, but I guess only Noaki knows how to do that.

--
GCP
Title: What should become of LAME?
Post by: JohnV on 2001-09-30 15:56:14
Well, MPC is already very highly tweaked. It'll take quite some time until Vorbis touches its quality at higher bitrates. Current MPC encoder will remain free, no matter what. Future decoders (which will open source, same as now) are able to decode current MPC-bitstream.

Andree hasn't said anything certain about the price. But he has said several times that most probably future encoders can be used for free with full quality. He has talked abot possible speed improvement for those who wish to support his work with small amount of cash.

Personally I like both projects: Vorbis and MPC. I prefer MPC at the moment because of the higher quality. In my opinion it doesn't mean I have to choose between these codecs now or maybe ever. I really see myself using both in the future. MPC decoder will remain open source.

Lame is not quite over yet, current version offers many people reasonable quality level. Some of the more fundamental features which could provide still essential qulity gain for Lame are at least:
- Use of mixed blocks (long blocks for the lower freqs and short blocks for the higher freqs)
- Nonlinear psychoacoustics model as described by F. Baumgarte (Naoki Shibata said he tested this, but it should be really examined.
- still improved noise shaping and block switching.

Unfortunately(?) it seems that most of the Lame developers think Lame starts to be "good enough".

i believe MP3's position will still remain strong for years.
Title: What should become of LAME?
Post by: David Nordin on 2001-09-30 17:33:41
I guess I just want to plan for the future..
shit I sound like a familyman planning for economy 
It's no fun now when I feel the death coming closer, I don't have the same enthusiasm.. 
I don't know if I should start thinking about moving over to vorbis or MPC now or what.. I'm confused:confused:
Title: What should become of LAME?
Post by: Dibrom on 2001-09-30 20:46:30
Most of the points JohnV raised are on target.

I don't really see where people get the idea that Buschel is going to be trying to make all kinds of money off of mpc, especially when it keeps being stated that this isn't likely the case at all.

There are a few issues I'd like to point out though.  One is that even if future versions where not available in some free manner (which I doubt will happen), the current version is so high quality that someone could be happy with it for a very long time.  Currently no other lossy codec is able to quite match it, though some AAC implementations are getting closer.  Second, if someone would be absolutely unwilling to pay $10 for a codec as good as mpc (so that the author could pay possible patent fees so that YOU could use this great codec), I actually think that's a little bit sad.

As for the rest.  I would like to see Vorbis get better and even become the best, but I'm not even sure its established at this point that it CAN get as good as mpc with its current specifications.  So far I am unconvinced.  I'm not saying it can't happen, but it hasn't yet, and I don't know if it can, at least until wavelets hit.  Unfortunately, wavelets won't be coming for a year or longer according to Monty.  So that's quite a ways to wait and see.  If you look at the current situation, AAC is really a pretty advanced codec which has had more time to mature than Vorbis and has the advantage of tried and true methodology on its side through the use of patented techniques, and even it doesn't quite match MPC's quality at higher bitrates.

I absolutely understand the point behind furthering a free open source codec, but when it comes to encoding my own music, I am more interested in what gives me quality NOW than what MIGHT give me similar quality in the future.  After all, the files I encode now aren't going to get better just because the codec may improve over time.

As for the majority of the LAME development team, if they do decide to migrate to another format in the future, I do think that Vorbis is probably the logical choice.  The people who I believe will be working on mpc in the near future, won't really have an effect on that.

Btw, I don't think LAME will really be going anywhere for awhile.  There still are a few things that can be improved upon that have been mentioned elsewhere such as the addition of the use of mixed blocks, a new tonality estimation function, better block switching overall, and a general speedup of much of the code behind nspsytune.
Title: What should become of LAME?
Post by: Jon Ingram on 2001-09-30 20:50:02
Quote
I don't know if I should start thinking about moving over to vorbis or MPC now or what.. I'm confused:confused:


Well, those that prefer MPC will tell you to use MPC, and those that prefer Vorbis will tell you to use Vorbis ...

(and I prefer Vorbis)

There is nothing terribly wrong with MP3. In fact, there is one thing majorly right with MP3 -- its hardware and software compatibility is better than anything else except uncompressed WAV files. So, if you are ripping audio to share it with friends that are still confused about why their computer comes with a cup holder, MP3 is the way to go.

Beyond that, which audio format you use depends on your tastes, and your needs. My needs make me use Vorbis, but yours may differ.

<see? I can do objective and balanced in short bursts...>
Title: What should become of LAME?
Post by: Gabriel on 2001-10-01 09:59:50
Well, I'm not sure if LAME is dead, but I'm sure that many of the developpers don't have as much time as before.
I think that the key is time. We are all lacking time as there are unfortunately only 24h per day.
If I should switch to another projet, it would be Ogg for sure.
But there are still things that can be worked on LAME:
*mixed blocks
*HQ filters
*IS
*better ath adjust
*improved m/s decision
*better tonality estimation
...

But all this need some time.
Title: What should become of LAME?
Post by: Milkman.dan on 2001-10-01 12:01:11
I too would rather see LAME dev. people going to Vorbis rather than another project.

Though, in all honesty, I hadn't even heard of MPC until a few days ago. I don't really know anything about it, how it works, or any of that.

What I do know is that its shareware, and the decoder is opensource. Which is good.

But my consumer heart tells me that Vorbis' truly opensource nature is going to be better in the long run...the real long run anyway.

As I understand it, MPC can't be beat for archival quality encodings. ( Of course, I'll have to hear that for myself ). But, and I'm making an assumption here,  if the encoder is controlled by one guy or a small team of people,  how can it keep its edge against such a large group of coders working on another project? I promise, I'm not trying to be arrogant or anything. I just don't understand.

How long will MPC's headstart last?
Can it be used safely in businesses without fear of recourse?
Is MPC's encoding engine/technology truly so much better in some fundamental way that Vorbis can't compete?

So please, lay it out for me...or point me somewhere so I can read up.

Thx.
Title: What should become of LAME?
Post by: Dibrom on 2001-10-01 12:25:59
Quote
Originally posted by Milkman.dan
I too would rather see LAME dev. people going to Vorbis rather than another project. 

Though, in all honesty, I hadn't even heard of MPC until a few days ago. I don't really know anything about it, how it works, or any of that. 

What I do know is that its shareware, and the decoder is open source. Which is good.


It's freeware, not shareware.  There *may* be a pay version in the future, but in one form or another there is going to be a free encoder. Whether or not it is the one currently available or another is not known yet.

Quote
 
As I understand it, MPC can't be beat for archival quality encodings. ( Of course, I'll have to hear that for myself ). But, and I'm making an assumption here,  if the encoder is controlled by one guy or a small team of people,  how can it keep its edge against such a large group of coders working on another project? I promise, I'm not trying to be arrogant or anything. I just don't understand.


Well I wonder this myself sometimes.  The fact of the matter is that mpc provides higher quality than any other lossy codec around currently.  Some codecs like AAC even have quite a bit of development behind them and at mid to high bitrates I still haven't heard an AAC encoder which can quite match the level of MPC.  All I can think of is that the author behind the codec is extremely talented and knows what he is doing.  Not to mention many of the people who help him test and tune the codec have some of the most sensitive hearing around.  When you are completely oriented around quality, that kind of thing helps immensely.

Quote
How long will MPC's headstart last?


I think this is the wrong question to ask.  I think instead the question should be "can other codecs match the quality of MPC with their current methodology?", not necessarily "when will it happen" or "how long can it stay ahead".

Quote
Can it be used safely in businesses without fear of recourse?


I'm not sure this has been established one way or another absolutely for sure yet.

Quote

Is MPC's encoding engine/technology truly so much better in some fundamental way that Vorbis can't compete?


I believe this is a hard question to answer.  Overall I think currently the fact that mpc does not have to rely on block switching like transform codecs do, gives it a huge advantage.  That's not to say that subband codecs do not have problems either, but it seems the author of mpc has done and extraordinarily good job of minimizing these to the point where they are no longer issues really.

I think Vorbis will probably be on much more equal footing when wavelets are implemented if that turns out to work as well as planned.

Quote
So please, lay it out for me...or point me somewhere so I can read up.


http://www.stud.uni-hannover.de/~andbusch/...ocoder_eng.html (http://www.stud.uni-hannover.de/~andbusch/audiocoder_eng.html)
Title: What should become of LAME?
Post by: Garf on 2001-10-01 16:43:49
Quote
Originally posted by Dibrom
quote:
--------------------------------------------------------------------------------
 
As I understand it, MPC can't be beat for archival quality encodings. ( Of course, I'll have to hear that for myself ). But, and I'm making an assumption here, if the encoder is controlled by one guy or a small team of people, how can it keep its edge against such a large group of coders working on another project? I promise, I'm not trying to be arrogant or anything. I just don't understand.
--------------------------------------------------------------------------------

Well I wonder this myself sometimes.  The fact of the matter is that mpc provides higher quality than any other lossy codec around currently.  Some codecs like AAC even have quite a bit of development behind them and at mid to high bitrates I still haven't heard an AAC encoder which can quite match the level of MPC.  All I can think of is that the author behind the codec is extremely talented and knows what he is doing.  Not to mention many of the people who help him test and tune the codec have some of the most sensitive hearing around.  When you are completely oriented around quality, that kind of thing helps immensely. 


Actually, the reasons aren't very complex.

The LAME project was essentially handicapped by the MP3 standard. You can have many good ideas, MP3 will only take you so far. MPC on the other hand extended MP2. Buschel could put in whatever he wanted and build on the known and tested technology (thats not to say he didn't invent any new...but he could use what was already there). Ogg could also put in whatever it wanted, but had to work its way around patents from the ground up, so they had to invent totally new stuff.

Effectively, MPC has both incompatibility (a good thing in this case) but it can also work with known and tested technology.

I think it's obvious this means a large headstart over competing projects, save perhaps AAC. But as you already explained, focussing on one specific area is also a great help.

For Ogg it's currently a bit unfortunate that Monty is the only one who can really work on the encoder. There are others, but they aren't as knowledged on the internals yet.

Quote
I believe this is a hard question to answer.  Overall I think currently the fact that mpc does not have to rely on block switching like transform codecs do, gives it a huge advantage.  That's not to say that subband codecs do not have problems either, but it seems the author of mpc has done and extraordinarily good job of minimizing these to the point where they are no longer issues really.


Actually, one generally expects transform codecs to do better. There is only problem: preecho. But it's a nasty one.

AAC has TNS, Vorbis will have wavelets. But even without them a transform codec can do quite well on preecho (see -dm-preset in LAME)

--
GCP
Title: What should become of LAME?
Post by: Jon Ingram on 2001-10-01 18:33:09
Quote
For Ogg it's currently a bit unfortunate that Monty is the only one who can really work on the encoder. There are others, but they aren't as knowledged on the internals yet. 


What we need is a nice abstract definition of the Vorbis 1 format, so that it would be possible to make a cleanroom implementation of a Vorbis encoder without having to get on intimate terms with the current one.

Given that I'm doing a PhD now, I keep on looking at Vorbis and thinking 'there's got to be a decent journal paper in this format for someone' :-).  Sadly it's a little out of my field (which is currently something weird to do with hyperbolic metric spaces and averaging phylogenetic trees).
Title: What should become of LAME?
Post by: Garf on 2001-10-02 01:02:28
Quote
Originally posted by Jon Ingram
Well, those that prefer MPC will tell you to use MPC, and those that prefer Vorbis will tell you to use Vorbis ...

(and I prefer Vorbis)


Pick whatever best suits your needs and feelings.

There

<advocacy>

I'll be the first to admit that MPC gives excellent quality and speed. In fact, as far as AQ is concerned, I think it's (by far) the best codec out there.

But I am also a Free Software zealot. So I do whatever is in my ability to make Vorbis just as good or better. I don't like the idea that MPC is patented and nonfree (and not even opensource on the encoder).

If a user doesn't care for the later issues, no amount of Vorbis advocacy is going to help, no matter how hard we scream or flame. The user will pick the solution that is best for him. If that's MPC, so be it. I am not going to _force_ people to use Free Software (unlike certain other people in the Free Software movement).

Rather, we should try to make Vorbis better than anything else out there to show how powerfull the Free Software idea is.

I'm a Free Software zealot because I have seen what the idea can do, I _know_ what it's done for me and I've grown to love it because of that. Arguing about RMS or how the GPL is viral isn't going to change my opinion about it.

Rant all you want, but what have _you_ done for _me_ ?

</end advocacy>

--
GCP
Title: What should become of LAME?
Post by: Garf on 2001-10-02 01:06:42
Quote
Originally posted by Jon Ingram
What we need is a nice abstract definition of the Vorbis 1 format, so that it would be possible to make a cleanroom implementation of a Vorbis encoder without having to get on intimate terms with the current one.


I doubt that would be of any use. You'll just end up reinventing the wheel all over again. Besides, Monty's code is some of the cleanest (well, usually...) I've seen.

The problem is simply that people with an intimate knowledge of cutting edge psychoacoustic audio compression and an infinite supply of ideas don't grow on trees.

--
GCP
Title: What should become of LAME?
Post by: Milkman.dan on 2001-10-02 10:10:36
Ah, thank you for the expansive response. I really appreciate it.


Quote

It's freeware, not shareware.  There *may* be a pay version in the future, but in one form or another there is going to be a free encoder. Whether or not it is the one currently available or another is not known yet.


oops. My mistake. Guess I didn't know after all.


Quote

Well I wonder this myself sometimes.  The fact of the matter is that mpc provides higher quality than any other lossy codec around currently.  Some codecs like AAC even have quite a bit of development behind them and at mid to high bitrates I still haven't heard an AAC encoder which can quite match the level of MPC.  All I can think of is that the author behind the codec is extremely talented and knows what he is doing.  Not to mention many of the people who help him test and tune the codec have some of the most sensitive hearing around.  When you are completely oriented around quality, that kind of thing helps immensely.


Well, I guess this just goes to the other question of whether MPC's encoding technique is just truly all-around better. If so, it should rise to dominance (or should. Beta anyone?)

I just have this sneaking suspicion that, because of Vorbis' BSD libraries and things of that nature, it makes commercial integration more appealing. (It certainly does in my case, anyway.) I just don't know enough about MPC to make a call on it. I'll fix that here in a minute when I go to that link you kindly provided.


Quote

I think this is the wrong question to ask.  I think instead the question should be "can other codecs match the quality of MPC with their current methodology?", not necessarily "when will it happen" or "how long can it stay ahead".


Hmm. Don't think I agree here. I think, personally, that the logistics of many people in dev. vs. less people weigh in here. Even if Ogg's technique is inferior, it may still supercede MPC just because of the effort expended in making it better per unit of time. Although I think you question is certainly relevent. Like I said, if MPC just can't be matched, well then, thats that.


Quote

I'm not sure this has been established one way or another absolutely for sure yet.

Alright. I'll see if there isn't a general question email I can use.

Quote

I believe this is a hard question to answer.  Overall I think currently the fact that mpc does not have to rely on block switching like transform codecs do, gives it a huge advantage.  That's not to say that subband codecs do not have problems either, but it seems the author of mpc has done and extraordinarily good job of minimizing these to the point where they are no longer issues really.

I think Vorbis will probably be on much more equal footing when wavelets are implemented if that turns out to work as well as planned.


Cool. I guess all I can say is that, if Ogg is at its current point using less-than-optimal tech, then that bodes well for the future.

Thanks for you time in reply.


http://www.stud.uni-hannover.de/~andbusch/...ocoder_eng.html (http://www.stud.uni-hannover.de/~andbusch/audiocoder_eng.html) [/quote]
Title: What should become of LAME?
Post by: Milkman.dan on 2001-10-02 10:14:22
Quote
I'm a Free Software zealot because I have seen what the idea can do, I _know_ what it's done for me and I've grown to love it because of that. Arguing about RMS or how the GPL is viral isn't going to change my opinion about it.

Rant all you want, but what have _you_ done for _me_ ?

</end advocacy>

-- 
GCP


w00t brotha. w00t.
Title: What should become of LAME?
Post by: Dibrom on 2001-10-02 10:43:44
Quote
Originally posted by Milkman.dan
Well, I guess this just goes to the other question of whether MPC's encoding technique is just truly all-around better. If so, it should rise to dominance (or should. Beta anyone?)


The sad truth is that technical superiority, and even higher quality, does not necessarily equate to mainstream acceptance.  I do believe MPC is gaining support, but I seriously doubt it will ever increase to a degree where it would be able to replace MP3.  I am actually counting on Vorbis to do that.

Quote
Hmm. Don't think I agree here. I think, personally, that the logistics of many people in dev. vs. less people weigh in here. Even if Ogg's technique is inferior, it may still supercede MPC just because of the effort expended in making it better per unit of time. Although I think you question is certainly relevant. Like I said, if MPC just can't be matched, well then, that's that.


This is exactly why I used AAC as an example.  There is already an example of many people (many with a vast amount of knowledge in the area of audio coding) developing a format vs this 1 person, and so far MPC still seems to be on top in terms of very high audio quality.

To be fair though, the focus of AAC and I believe even Vorbis probably, is to provide adequate quality at a very acceptable bitrate, not necessarily transparent quality with little regard to bitrate (which is I think more inline with what MPC attempts to achieve).  Some may argue this vehemently, but the fact of the matter is, current development in both formats focuses far more on lower bitrates with a more averaged bitrate approach.

Quote
Cool. I guess all I can say is that, if Ogg is at its current point using less-than-optimal tech, then that bodes well for the future.


Well I think "less than optimal" may be a bit harsh, though probably technically accurate.  I'd probably say instead that "it could be better".  However, it still hasn't been clearly established that wavelets will actually make the difference expected.  Theoretically if used correctly they should provide improvements, but as is shown all too often, improvements in theory and actual improvements after implementation, do not always quite match up.  I think we will have to just wait and see, but I do agree that there is quite a bit of promise for future improvement.

Quote
Thanks for you time in reply.


No Problem
Title: What should become of LAME?
Post by: Garf on 2001-10-02 12:26:23
Quote
Originally posted by Milkman.dan

w00t brotha. w00t.


The 'advocacy' warning wasn't there for nothing

--
GCP
Title: What should become of LAME?
Post by: Milkman.dan on 2001-10-03 09:04:39
Yeah, I would have to agree that Vorbis and AAC, as compared to MPC, have different focuses. And that certainly shows in development.

I think, to a certain extent, Vorbis addresses the transparent application with the lossless channel mode that is talked about in one of the white papers on the site. Of course, since it isn't implemented yet(or if it is, I have no idea how to access it), we can't really tell what the effect on bitrate/filesize will be.  Any thoughts on that? If Ogg has a lossless or near-lossless profile availible, would that make a difference to you personally?

Perhaps that's a bit ambiguous. How about this:
Would the opensource nature of Vorbis( and thus its high adoptability), combined with a high-quality profile, change your mind about the standing of the codec in the audiophile world?

Personally, I look at it this way. I have pretty solid hearing. I treat my ears well, and for the most part, I hear flaws in all sorts on encoded songs that I, and others have done. And since I have two applications for music on my computer, I have two different sets on needs.

On one hand, I have gobs of hard disk space availible to me. So, 90% of the time, I don't really care how much space my music collection takes up. I don't want it to be excessive, but I don't mind giving up the space if it is well used.

Application two is portable audio. I have a Rio500, and I get about 45min of 160kbpsVBR on the thing. And it's tolerable to listen to on the go. But it's not very high quality.

From my viewpoint, Anything that makes it easier for companies to integrate additional codecs, is better for me. Vorbis has the edge here. And thanks to the ripple effect, and a healthy dose of lazy, I don't want to have to transcode something to use it on a portable. So if I have the same files on both applications, that's great for me.

That's why I tend to stick up for Ogg. A lossless mode, or a near-lossless profile would be great. But for me, 192kbps ought to do it, assuming they get all their ducks in a row. I would guess that bitrate would be the middle ground between the two apps.

Anyway, sorry for rambling on. Just wanted to talk a bit. I'm interested to see what you'll say Dibrom. Or anyone else, for that matter.

Later.
Title: What should become of LAME?
Post by: ChristianHJW on 2001-10-08 09:57:27
Let me ask for all DVD rippers ( DivX ) :

which codec has the best chances to outperform MP3 for bitrates < 128 kbps ? I am using Lame ABR now as recommended by R3MIX forum members.

Is MPC prepared and designed for lower bitrates ?

Ogg Vorbis the one that will perform best at these bitrates ?

Or is AAC ( Nic could really use somebody to help him here ) the future for DivXers ?
Title: What should become of LAME?
Post by: Garf on 2001-10-08 12:05:31
Quote
Originally posted by Milkman.dan

I think, to a certain extent, Vorbis addresses the transparent application with the lossless channel mode that is talked about in one of the white papers on the site. Of course, since it isn't implemented yet(or if it is, I have no idea how to access it), we can't really tell what the effect on bitrate/filesize will be.  Any thoughts on that? 


The lossless mode just relates to channel coupling. It is comparable to mid-side stereo in MP3, but it's called lossless because unlike MP3 the transformation from L/R to M/S never loses a bit of precision.

This doesn't mean the codec itself gets lossless. Far from it.

It is already implemented (actually, it has always been because the pre-RC2 dual stereo is also lossless). You don't have to activate it. The encoder decides whether to use it or not depending on the audio content that you are encoding and the quality you requested. Bitrates from 256kbps and up always use lossless stereo, and so does my tuned Vorbis mode.

RC2 had some quality problems with the non-lossless modes, but they are (will be) corrected in later versions, and hence it's possible lossless coupling will be used less often in the future.

I believe the difference between 128kbps RC2 channel coupling and lossless coupling is on the order of 10-20kbps.

Quote
If Ogg has a lossless or near-lossless profile availible, would that make a difference to you personally?[/b]


Sure does. That's why I'm working on tuning high-bitrate Vorbis. The next version (RC3) should already have much improved high-bitrate modes.

A lossless mode would be nice too, but that won't be part of Vorbis (although part of Ogg, the bitstream format). There is going to come one after Vorbis stabilizes a bit more, based on Squish,FLAC and possibly Monkey's Audio.

--
GCP
Title: What should become of LAME?
Post by: Garf on 2001-10-08 12:11:35
Quote
Originally posted by ChristianHJW

Is MPC prepared and designed for lower bitrates ?


No. MPC is expressly designed as a high-quality high-bitrate format. It's advantage over Vorbis/AAC/MP3 at high bitrates (no frequency domain transform) becomes a large handicap at lower bitrates.

Quote

Ogg Vorbis the one that will perform best at these bitrates ?


That's what we hope. It should do pretty well right now, and it'll only get better.

Quote
Or is AAC ( Nic could really use somebody to help him here ) the future for DivXers ?


Given the many restrictions on that format, I'd hope not. Playing AAC movies on my linux machines would be problematic to say the very least.

Qualitywise it should do pretty ok, given it's designed for 96/128kbps, but I think Ogg Vorbis can be at least equal or even better, if not now, then certainly in the future.

--
GCP
Title: What should become of LAME?
Post by: Milkman.dan on 2001-10-09 12:35:37
Well, maybe there should be an express lossless mode for Ogg? Like HuffYUV the video compressor? I use Huff all the time, and it is a wonderful piece of work. It would be nice to be able to capture raw video footage with compressed video and audio, both being very flexible and opensource.
Title: What should become of LAME?
Post by: YouriP on 2001-10-09 14:40:12
Agreed. If we could use Ogg (given it will use the compression features of the programs mentioned) to record audio losslessly, that would be a blessing. Of course, we'd need it to be ACM for use with capturing programs like VirtualDub.

Btw, hasn't the author of LPAC any intentions of releasing the source for his program?
Title: What should become of LAME?
Post by: Jon Ingram on 2001-10-09 18:36:41
Quote
Btw, hasn't the author of LPAC any intentions of releasing the source for his program?


No need to use LPAC, there is already an excellent and relatively fast open source lossless encoder in FLAC:

http://flac.sourceforge.net/ (http://flac.sourceforge.net/)

There is a comparison chart there which shows FLAC performing a tiny bit worse than LPAC but no slower. Monkeys Audio is faster (in some modes) and has better compression (about 2.5% more) than both, but it's completely unusable as an archive format (no specification, no source, Windows only).

There was a nascent project to strip FLAC of its inbuilt streaming wrapper and use the Ogg wrapping format instead -- I've seen interest in mailing lists recently about this.
Title: What should become of LAME?
Post by: YouriP on 2001-10-09 18:58:54
If the author of LPAC should release the sourcecode to his program (same for Monkey's Audio - which Matt actually said he would at some point) then those can be used to merge them all together into one lossless, patentless and free codec in Ogg. It's not like the exact code for each has to be used, just portions of it or certain algorithms to create the Mother of All lossless encoders.
Title: What should become of LAME?
Post by: ChristianHJW on 2001-10-09 21:41:36
@Garf :

Thanks for your clear answer. So i better go assist ingoralfblume on getting Ogg to work in AVIs without sync problems instead of disturbing your hi-level developers talk  ....
Title: What should become of LAME?
Post by: Jon Ingram on 2001-10-09 22:56:12
Quote
The Mother of All lossless encoders

I like the sound of that... what would the logo be? Perhaps a happy and munificent bust of Saddam Hussein...
Title: What should become of LAME?
Post by: Gabriel on 2001-11-07 08:05:45
There is loseless audio coding in the Ogg suite. But it's not Vorbis, it's Squish
Title: What should become of LAME?
Post by: Garf on 2001-11-07 09:02:48
Quote
Originally posted by Gabriel
There is loseless audio coding in the Ogg suite. But it's not Vorbis, it's Squish


Squish hasn't been developed for ages and does not support the Ogg bitstream format.

FLAC is, and has support for Ogg in the latest versions.

--
GCP
SimplePortal 1.0.0 RC1 © 2008-2021