Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: My impressions on the current state of codec devel (Read 29158 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

My impressions on the current state of codec devel

Seems that Lame and many other audio codecs development is a bit stalled, AAC excluded. Only AAC codecs seems to be in fast development phase atm.

However, there has been a rival (for Lame) MP3 encoder in works somewhat secretly by a known person. Don't know the latest details at this moment, but could be that it is in alpha stage soon. And I'm not talking about Lame 4.0.

Anyway, imo it could be interesting if more free development resources were pointed at MPEG AAC encoders, which offer potential for clearly higher quality than MP3 and are ISO standard compliant. At this moment I don't see so great advancement in the spreading of Vorbis, while AAC news keep coming in at increasing rate.

Also it's a deficiency, that there's practically only one Vorbis encoder: There's no Vorbis encoder competition except cross-format, where as there are several different AAC-encoders competing.
Seems that it's not enough to drive Vorbis development further: there's just not enough resources at Xiph.
Also the Vorbis hardware support keeps delaying and delaying...

Seems that MPC development is going very slow too. At least there's not much news from Frank, except some interesting (private) thoughts by him that he's considering going MDCT using 3 different block sizes for SV8 or maybe SV9.

Hmm.. this message was actually in another thread, but I split it and created a new one. Wanted to write something a bit more spicy for a change. 
Juha Laaksonheimo

My impressions on the current state of codec devel

Reply #1 –
A new MP3 codec? IMHO most people agree that MP3 quality peaked with lame 3.90.3 and only some minor lower bitrate tuning could be done, if people felt there still was the need for it (obviously some do).

But is there still need for it? AAC is spreading faster than most people (including myself) ever imagined after seeing Vorbis not moving forward for almost a year now. Maybe there should be put more effort in existing (open-source) implementations like FAAC to keep them up with the commercial level.

dev0
"To understand me, you'll have to swallow a world." Or maybe your words.

My impressions on the current state of codec devel

Reply #2 –
Hmm nice thread... 

mp3: Developers work on what they wish, no offense. But isn't mp3 getting obsolete? Any further tweaks, modifications won't bring too much quality. Why not focus the effort onto other projects?

Vorbis: Unfortunately Vorbis project reminds me of those stocks that look very promising at first and then gets delisted and sold at 1 cents a share. I was interested in one such company: Constellation 3D developing FMD discs before they went dead  Anyway, Vorbis moves real slow, and seems like MP4 will be there before them.

AAC/MP4: Shows quite a promise, especially with Apple's adoption. Will probably be ubiquitous in a couple of years.

MPC: Rocks as it is. Will probably never become popular, and that's actually quite good . We won't have crappy MPC encoders around or people claiming they encode their music into mathematically loseless MPC. I for one would be quite upset if there won't be any lossless SV7->SV8 converters, or SV8 quality turns out so damn good compared to SV7 to make me reconsider reripping. But at least on MPC ground work is not fragmented as in the Vorbis project (Theora,...) and so it seems like it's moving faster  Hardware support will inevitably come when we could program our own decoders for our players. They're getting more complex every day. Soon one could run an OS on their player.
The object of mankind lies in its highest individuals.
One must have chaos in oneself to be able to give birth to a dancing star.

My impressions on the current state of codec devel

Reply #3 –
Very interesting thread, really.

I was happy to read on the Vorbis-dev ML today that there is serious work being done on a working 5.1 Vorbis implementation. Of course, they wouldnt be Xiph if they would just take what is proven to good and wide spread, no they have to redo everything and make a new 5.0 standard based on an ambisonics approach, making it almost impossible to simply transcode a 5.1 DTS or AC3 sound stream into a Vorbis file, without having to to modify/remix the individual channels on a PCM basis, with complex algo's, before encoding, but thats fine with me ... it will be the perfect way for DSPguru to prove his unbeatable Digital Processing knowledge, and a good reason to reactivate him from pensioneering  !!

About the slow Vorbis development recently : well, the upcoming Theora was certainly the reason for this, and i still have doubts if it was a wise move to set on the old ON2 VP3 video codec, instead of improving Vorbis and pushing Tarkin as a parallel development same time. The decision to make Theora had a huge demotivating factor for the Tarkin developers it seems, both the MLs seem to be dead since a couple of months, all devs seem to have left.

In the end, and there can be no doubt about that, once Theora is here and Xiph people manage to bring some good codec developers in, maybe even from the FFMPEG or XviD scene ( like Mike Melanson or Michael Nidermayer ), so they can offer better quality at low bitrates than VP3 does currently offer, than there will be a big run towards Theora as a patent and license free video streaming solution, and this will hopefully have a positive effect on Vorbis development also ... we'll see !

After almost one year of fighting with the Ogg bitstream adaption to fit for video needs, Monty for sure will have a couple of very good ideas on how to tweal Vorbis, and being the real full hearted codec developer he is, he will be glad to have the chance to make them happen, no doubt  !


About AAC :

Its doing extremely well currently, and we are proud to be amongst the first video container formats to support it in matroska, and since the latest VirtualdubMod release we can even edit/cut matroska movies with AAC audio ( not even the highly official MP4 can do this yet  ! ), and since Microsoft released a WMV9 VCM codec recently, we can finally even use this very nice video codec with some good sounding audio, instead of crappy WMA  .... my experiences with Nero's latest AAC encoder are so extremely positive that i decided to transcode all AC3 to AAC 5.1 now, and add both the English and the German audio track to my typical 2 CD backups, instead of keeping the German AC3 and adding a English Vorbis Stereo track at q 0.2 to it  ......

AAC's future is certain, it will definitely become the worlds next audio compression standard, with wide spread hardware support, the real MP3 successor. But my heart, please forgive me all you talented people working at Ahead and elsewhere, is still pounding for Vorbis ... maybe i am in the OSS scene too long already  ....

My impressions on the current state of codec devel

Reply #4 –
A nice teasing thread....


...and it works, as I am now wondering about who...

My impressions on the current state of codec devel

Reply #5 –
Quote
A nice teasing thread....


...and it works, as I am now wondering about who...

No, I'm of course NOT talking about MyMe/Hegemann, but somebody else (and you might guess who it is, although you first might think "can it be him out of the dark.." ) who at least last time I spoke to him some weeks ago was targetting that the codec outputs mp3-stream soon, so that we could go officially public. 

Anyway, don't Lame devs have any interest in starting an AAC codec or attend the FAAC project?
Juha Laaksonheimo

My impressions on the current state of codec devel

Reply #6 –
As i know a lot of users of mp3's, i know that the jo public won't adopt AAC as a standard because they're not familiar with the name. IMHO, i believe that AAC will get a wide spread recognition it it was referred to more as MP4 rather than the common AAC naming... People will look at the name MP4 and think "ooh, its the next version of the mp3, and it must be better" and will adopt it.

Well thats just my 2p worth.... 

Edit: Reworded it to make more sense 

My impressions on the current state of codec devel

Reply #7 –
I personally would actually like to see more MP3 development, because it's a pretty entrenched DRM-less format: companies are stuck supporting it whether they like it or not, and I'd like to keep things that way (vs. only have WMA/AAC support with DRM stuff).  In fact I think the only major reason companies are moving to AAC is the DRM; if they cared about quality they would've done stuff like LAME's vbr years ago instead of sticking to crappy 128 CBR mp3s.

I think MP3 actually is "good enough" as a format for indefinite use, especially for the 99% of people who aren't HA.org users.  I also don't think it's true that LAME is currently the best MP3 can get: it's the best LAME can get without a major rewrite, but that's not the same thing.  There are major deficiencies in both the GPSYCHO and npsytune models that the --presets have to patch over, often wasting bits in an attempt to catch some corner cases.  I think with a much more ideal psymodel --preset standard like performance but ~20kbps lower, or possibly a bit more, should be possible (and hopefully with fewer killer samples).  And with hard drives getting bigger all the time, "almost always transparent at 160-180kbps" is probably good enough, especially if a well-performing --preset medium at around 128kbps is possible with a new psymodel (GPSYCHO and npsytune currently seem to fail especially badly at the lower bitrates, sometimes even doing worse than ABR).  That would make it indistinguishable from AAC 128kbps for the majority of people (yeah, some of you here could probably ABX it, but you'd be using --preset standard anyway).

My impressions on the current state of codec devel

Reply #8 –
i'm very disapointed by vorbis. i really thought this format has future when 1.0 came out because it was better than mp3 and open source and had chances to get hardware support. actually nothing happened since 1.0. no hardware support, no final version which is tuned for high bitrates (-q6). nothing.

then i can also use mpc. because it has better quality. i mean ok, it will never get hardware support but vorbis doesn't seem to get it very soon too. or i'll switch to aac which gets more and more hardware support and there's at least some development!

my impression is that aac will be the next standard because it gets more and more hardware support and people will use it because of the name mp4. mp3 => mp4... thats like windows 98 and windows 2000. newer = better. if vorbis doesn't offer better quality in near future or gets hardware support acc will win the race and vorbis will be dead. (my opinion). no arguments to use it anymore

My impressions on the current state of codec devel

Reply #9 –
yeah its been almost a year since vorbis 1.0..  amazing so little has happened with it since, except for internal squabbles and hype for portables that are always 'readying' their vorbis firmware for release, but somehow they never quite finish it.

My impressions on the current state of codec devel

Reply #10 –
IMHO, if you think that mp3 is nearing the end of the line as a useful codec, think again. For mp3 to be replaced in terms of popularity and usage will require the succeeding codec to be distributed as freely as mp3 is now and for mp3 to be withdrawn.

Let's face it, the vast majority of people busily encoding away probably don't even appreciate that there is a choice of mp3 codecs available, they just fire up MusicMatch, or whatever, and away they go. For most of these people, their mp3 tracks are played in environments where peak audio quality is almost irrelevant since the background noise is too overpowering. And, in reality, if they are just being used to provide background music, quality is hardly critical. It is probably arguable as to whether any lossy codec should be considered at all for audiophile use.

Anyway, enough of that. In my view, a new mp3 codec can only be good, especially if it proves to be superior to what is currently available. If my sources are correct, I believe that this new encoder uses a psy model based on the Musepack model. Now that has to be something worth giving some listening time to, doesn't it?

But we'll have to wait and see......

My impressions on the current state of codec devel

Reply #11 –
Quote
I believe that this new encoder uses a psy model based on the Musepack model.

Hehe.. Rumors rumors, from rjamorim if I'm not mistaken. 
Anyway, I don't think it will be so. At least not entirely. The person is improving and developing his own work.

But I have to say that I don't have any very recent update about the situation of this encoder.
Juha Laaksonheimo

My impressions on the current state of codec devel

Reply #12 –
Quote
Hehe.. Rumors rumors, from rjamorim if I'm not mistaken.

And there was me trying to be discrete!!!!

My impressions on the current state of codec devel

Reply #13 –
Can we get a name of the new encoder?

My impressions on the current state of codec devel

Reply #14 –
Quote
Can we get a name of the new encoder?

Not at this point, since it's unclear how the project is advancing. I think when it outputs something the idea was to go public then. Although it seems that rumors have been spreading for some time already.
Juha Laaksonheimo

My impressions on the current state of codec devel

Reply #15 –
Actually I think vorbis developement is going great - I know there haven't been many updates but to me that's a pro.  Vorbis already works very well, and I'm not that interested in arguable quality differences between ogg/mpc/aac so long as excellent quality is achieved.  I can generally hardly hear or not hear at all the difference between vorbis at q5 and the original for most sampled on a headphone (that's not to say it isn't easy for some samples), and as I'm most interested in playing over speakers (making the difference inaudible) this is fine for me.

Vorbis developement is going just fine.  They have a stable, working, good release and shouldn't release every few weeks unless necessary - but the quality differences certainly wouldn't make that necessary (as I said vorbis is near transparent at q5 anyhow).  There should be a new vorbis release later this year, which is fast enough.  And vorbis isn't standing still; it's the codec of choice for many games (and probably other software packages too), because it is very high quality (no other codec is _significantly_ better), stable (no new releases every time you turn around), flexible/simple (it'll handle low and high bitrates in one package), and safe (it's open source with a very lenient license for the decoding lib, meaning you don't have to worry about future incompatibility).

Mpc is much more risky in most aspects; it's smaller and the code isn't open (so who knows how long it'll be around),  there are patent issues to contend with (might get killed or be in legal trouble some day), it's only useful for high bitrates, the software support libraries are much less extensive.

Then there is AAC - a threat to vorbis because of marketing muscle.  It would never win on quality, simple because both are sufficiently good for that not to be an issue, and with so many AAC implementations around, it's rather a confusing place.  AAC is patented and not easily available therefore.  It has DRM issues.  AAC is only interesting if it's pushed by the idustry (such as apple is doing), but it's not clear whether it will ever replace mp3 - I think it's unlikely because of it's restrictive nature, which is a hassle for both users and programmers.

Seriously - is there any reason to want vorbis developement to accelerate?  If vorbis were putting out versions constantly the software industry uptake would probably only be lower...

--Eamon

My impressions on the current state of codec devel

Reply #16 –
Quote
It has DRM issues.

DRM issues?

You mean the same issue which is also available to be used with Vorbis, the Media-S (formerly known as OGG-S) project?
http://www.sidespace.com/products/oggs/

DRM can be used or not, either with MPEG4 AAC or Vorbis.

Open Source DRM project was created because there obviously was a need for it...

Anyway, everybody can notice that Vorbis is not really advancing as fast as it was hoped. There's not any mass-user support, practically none hardware support etc. Some games have used it during the time it has been out. But not much any announcement about new industry deals etc.

But as one friend said to me, industry likes standards, and companies want to be absolutely sure that there are no patent issues which could suddenly appear. MPEG4 is at least clear regarding this: You pay licence fees and sleep well.. Anyway, I'd like to know where I can find the documentation of the alleged thorough patent search done by AOL? I'd like to know was the patent search as thorough as Xiph wants to make it sound. It just sounds weird that AOL is willing to spend lots of money to include some winamp pluging which can be downloaded easily separately, especially since AOL seems to be much more interested in other formats than Vorbis.  If there was a conflict, AOL could easily drop the Vorbis plugin.
Also it doesn't help that FhG personnel keep whispering at least privately in industry happenings that there are patent issues in Vorbis. It certainly makes companies think.. maybe better pay little to be safe than sorry.
Juha Laaksonheimo

My impressions on the current state of codec devel

Reply #17 –
Quote
Seriously - is there any reason to want vorbis developement to accelerate?  If vorbis were putting out versions constantly the software industry uptake would probably only be lower...


No good multichannel (>2) support is a significant handicap compared to AAC.

Quote
Mpc is much more risky in most aspects; it's smaller and the code isn't open (so who knows how long it'll be around), 


False

Quote
there are patent issues to contend with (might get killed or be in legal trouble some day)


Not much more so than Vorbis IMHO.

Quote
it's only useful for high bitrates


It does very well in the mid and lower mid range too now.

Quote
the software support libraries are much less extensive.


What is needed is there. There's no extensive framing support like ogg has and no readymade seperated encoder lib, but is that a problem?

My impressions on the current state of codec devel

Reply #18 –
Quote
yeah its been almost a year since vorbis 1.0..  amazing so little has happened with it since, except for internal squabbles and hype for portables that are always 'readying' their vorbis firmware for release, but somehow they never quite finish it.

Well, as long as those companies are paying the Vorbis developers

Right now, a significant part of the team works on Neuros (main reason why the core libs are not advancing). I suppose the improvements in the decoder (faster, less RAM), will be opened as well, which benefits everyone. (Regardless of what happens to the neuros itself)

My impressions on the current state of codec devel

Reply #19 –
Mmmm... he heh! *smirk* This is good news indeed! In fact, the most exciting news in a looong time! Kiitos JohnV!

Quote
I also don't think it's true that LAME is currently the best MP3 can get: it's the best LAME can get without a major rewrite, but that's not the same thing.


I agree. It seems like LAME has ended up in a dead end as it is. A major re-write of the code or a totally new encoder is needed to significantly improve the quality. LAME 4.0 looks promising though.

My impressions on the current state of codec devel

Reply #20 –
Quote
Anyway, don't Lame devs have any interest in starting an AAC codec or attend the FAAC project?


I think that most of us do not have much spare time....

My impressions on the current state of codec devel

Reply #21 –
Yeah, its mp55

Ruairi

P.S. I'm joking!! I'm not writing a codec!
rc55.com - nothing going on

My impressions on the current state of codec devel

Reply #22 –
I think mp3´s time isn´t over yet, but I don´t think there will be lot´s of improvements in the future. Even if it reaches alt-preset standard quality at 128kbps, that isn´t what the most users want. A portable player with 64MB memory that plays WMA and MP3, most people will use WMA instead of MP3 (maybe even reencode mp3s from the internet), because it needs less space and on the crappy earphones it will sound nearly the same. Additionally MP3 has very much software support, nearly every audio program supports mp3 by default.
Ok, Vorbis is supported too by much programs, but only through plug-ins, like for CoolEdit, Nero or WMP.
Many people don´t like downloding things from the internet if it already works (hey, why do I need Vorbis, it works fine with mp3, maybe I get a virus or my system won´t work after installing that (never change a running system)). In my opinion, for example Xiph should do the following:
Develope a plug-in for an good audio-program, like Cool Edit (there is alread one, maybe they update it a bit), then sent it to Styrillium (or whatever this name is spelled) with the words like
"we have developed a Vorbis Codec plug-in for your great audio editing program "Cool Edit". If you like you can include and sell this fully functional plug-in for our audio codec with your applications. You don´t need to pay anything, nor subscribe something. Ogg Vorbis is completly patent-free, there are no licensing issues. Including this plug-in is good for Styrillium and for Xiphophorus. You can say Cool Edit supports that format by default and more probably more users will buy it. On the other hand with such a good program supporting Ogg Vorbis, Ogg Vorbis will get used more widely."
Of course not only to Styrillium, to many widly used audio software producers.
On the other hand, they have to tune Vorbis more and more, compared to "Windows Media Audio 9" "Vorbis 1.0" sounds really young (also they brought out much more versions than M$). Multichannel audio has to be improved, they have to advert it "Vorbis 1.1, now with improved quality on low and high bitrates and improved multichannel (5.1/6.1/7.1 sound) support!"
If many applications support it by default, users thinks it is better than mp3 and portable support - More users will use it. Yust look at WMA.

For WMA, what is bad about WMA? The codec itself or M$ implementation? There is an open-source decoder for it, so it could be possible to write an encoder for it, but could it be better than M$ ones, or is the format so bad that is simply not possible to reach higher quality without breaking compatibility?

For Lame4 and "the new mp3 encoder". I think that is the wrong direction, there are already better alternatives to mp3, why is programming an encoder for and old standard? Why aren´t they developing a new/improved AAC, WMA (if possible) or Vorbis encoder? This are 3 formats where I can say they could be the future and widely supported, but for 2 of them there are no good free encoders avaible and for Vorbis there is also much that could be done.

My impressions on the current state of codec devel

Reply #23 –
Quote
I think that most of us do not have much spare time....

I suppose that's the problem for most codec developers as there isn't any commercial benefit for creating something like LAME unless you're lucky like Ivan

An example I would like to see is say make Lame 4 the final version - after any bugfixes or minor tweaks, the devs spent what time the had for LAME on FAAC..  it would be opening a whole new challange for some gifted minds

Imagine freeware AAC that whaled on every commercial implimentation (as is the case with mp3?)  W00t
< w o g o n e . c o m / l o l >

My impressions on the current state of codec devel

Reply #24 –
Quote
Anyway, don't Lame devs have any interest in starting an AAC codec or attend the FAAC project?

yeah that would be great!
I know, that I know nothing (Socrates)