ogg mux cannot handle vbr mp3 ?!?!?

[Sven Bent]

i whenever i tried to mux an avi+mp3->.ogm
i have no sund
wmp 6.4 and wmp8 wants to download the layer 3 codec again. but still no sound

if i have muxed avi+mp3->.avi with nandub and then avi -> .ogm with ogmux
the sound is ok...but no subs

:-(

Im using
winxp pro
oggmux 7.6
oggDS 0.990

[Koepi]

when you read the tooltip to the "movie file", then you'd know the avi _mustn't_ contain any soundtracks.
It won't add subtitles or other tracks then.

I didn't test vbr mp3. What do you need it for when you have vorbis?

*gna*

Thanks for the test.

Regards,
Koepi

[Sven Bent]

Originally posted by Koepi

I didn't test vbr mp3. What do you need it for when you have vorbis?



cuase i haven stil yet to see ogg out performan mp3 with MY ears.

ogg have BIG (and i MEAN big) pre-echo artifacts in test sampels like spahm/fatboy

the other test samples i can't hardly heart a big diffference.
so i would rather go with some small quality loss the risc ending up with som BIG GIGANTIC pre.echo artifact that even my dead grandfather would hear.

that why i use vbr mp3....until AAC is ready :-)


btw seeking is also screewed (no video) with muxin vbr mp3 with oggmux. and some VERY big/small percentage output when almost done muxin files (18762% done)

[TobiasWaldvogel]

The Ogg multiplexer relies on the timecode from the preceeding filter. Unfortunately the MPEG splitter sends only non-sense data.

If the MP3 track is already in an AVI it should work. Try to construct the corresponding graph in graphedit to check if it works.

BTW: I'm still using the RC3 vorbis DLLs and I'm very satisfied with the result. Well, I don't compare it with anything else. I just want to have sound tracks that sound good to me and that have a sutable size. It doesn't matter if it is technically more or less different to the orignal.

Is there really someone out there who watches 5 times the same movie to check if it is really exactly the same?

On the other hand I hear a REALLY big difference between 80kbs Vorbis and MP3

[Koepi]

I can't confirm pre echo problems with vorbis.
Must be something with your setup. Or you tested ogg vorbis beta 1 or something(even there the pre echo prblems from > 1 1/2 years ago are gone).

Koepi

[ChristianHJW]

.. well, there was a small problem with pre-echo in recent CVS builds .. Monty screwed something :D ... but these are fixed now, everything fine again ...

[Sven Bent]

I can't confirm pre echo problems with vorbis.

just try and download RC3 and the test sample spahm.wav or fatboy

Spahm:
ogg at 160 is evne worse then aac or mp3-aps at 128kbits
ogg.vorbis is RELLY screwed at this sample.

fatboy:
ogg at 185kbits soud more like mp3-aps 147
AAC outperforms both much at 148kbits


besids it really hard to hit a certan bitrates with vorbis without doing abr. and abr should degrade quality much with ogg (ab not really tuned yet)

also RC4 are on the steps.. It should have better quality for low bitrates and the pre-echo artifact gone


Note that these test samples are hard to encode test samples and are sledom produced in music..even thougt fatboy comes from a music track (don't know about spahm)

this will be the last post about mp3 vs vorbis else we just have to move the thread to audio codin area :eek:

[Koepi]

Nope, not really the last post.

I know those samples, I know the whole "hard to code test samples" library.

If you'd please read christian's post again, you'd see that the screwed up code is gone again.

Don't mix it all up and make people think vorbis is bad.
It's better than mp3 in any configuration, it's only beaten by mpc and *scratching head* forgot about the other format.

Koepi

[MaTTeR]

I have to weigh in here also and have to agree with Koepi on two points.

Latest CVS has cleaned up the pre echo and logical bugs AFAIK. This was discussed extensively in #vorbis and #project_mayhem IRC channels.

Secondly, IMHO Vorbis has been embraced by welcome arms here in the community for a number of reasons. One of the biggest reasons is that 9 out of 10 movies will have smaller file sizes and 98% of us love the quality it provides. Lets face it, our movies that we transcode to Vorbis isn't exactly classical music where we need to hear the crisp notes of a harp in the background:)

The exodus in this community has already started long ago, we no longer need MP3:D

[Nic]

I know Sven is a bit strong in opinion.....but I do have to back him up.

Sven Bent has been in audio as long (longer?) than I have & has always had problems with Ogg, he really finds the pre-echo irratating. You'd be surprised how, if you listen for artifacts too much, you become very sensitive to a certain type.

I tried to persuade Sven to like Ogg more a long time ago, but he really doesn't like the sound....I too prefer AAC to Ogg.
Also RC2, IMHO, handled the 64kbit region far better than RC3 (although I hear that problem is getting better all the time & may now be completely fixed :) )

But in the video encoding world, we like low bitrates....& Ogg is better than AAC at that (at present..., in reality, as LiquidAudio has shown AAC is also very good at low bitrates, however Psytel is tuned more for audiophiles :) )

-Nic

ps
For anyone that wants the samples for testing:
http://lame.sourceforge.net/download/samples/

[101]

Originally posted by Sven Bent

cuase i haven stil yet to see ogg out performan mp3 with MY ears.


mp3 or vorbis for movies is not a question... and each word is important in the previous sentence!

101.

[Sven Bent]

first let me get this clear i NEVER said OR meant ogg was better then mp3 on a geneeral basis. or that peopel should prefer mp3 over ogg.

if you (all of you who are readin) would look in the audio encoding forum you woudl se that i recommmend ogg.vorbis over mp3-aps.

i do beleive that ogg are doing a better job on 95% percvent of the materiel out there if not more. but RC3 from the the vorbis ebsite DUE have some probelms with pre-echo


Latest CVS has cleaned up the pre echo and logical bugs AFAIK. This was discussed extensively in #vorbis and #project_mayhem IRC channels.

I'm bot going to argue about cvs or other stuff or anything about the internal way ogg works.. mostly because i dont understand it ( i dont even knwo what cvs mean, but i get a faint idea) :)

but the test tells from the self. if you woulde to argue against me the ogg does NOT have problems at fatboy/spahm test sample.
downloed these and test for you self
compare agains psytell AAC and mp3-aps

was i meant whas that ogg are probaly duing 10% better qualtiy in 95 but ihaw a 40% lesser quality then mp3 in the last 5% (fatboyt and spahm and the alike). i woulde rather live with the 10% loss on all my soundtrakc then risck that one sound track for even a short perido was was loosing the 60% qualtiy drop ogg has over mp3 (in these VERY rare situations)

Hope i could say this the right way i meant it.
not haivn englsih as my native tounge are a bit difficult when explaining liek this.


BUT this thread was NOT about ogg VS mp3. so PLEASE lets us not "fight" this or at least "figth" the proper thread in the propper fora... audio coding.

my thread was just a statemen of oggmux not muxin mp3 vbr into .ogm
so that i could either

1: Get it confirmed
2: Know what i do wrong
3: Might have koepi to fix this issues.. if it at all is an issues with oggmux and not the ogg container itself



wohoo what a fuzz :)
anyway
please bear over with me for spelling errors it is late and i will correct them later

[ShinerSpoon]

I also noticed the mp3 option in the later versions of oggmux and figure that Koepi had just got aggravated with silly questions and decided to make a universal muxer for most all audio types and divx. I confirm it does not playback in mp6.4 while the same avi and mp3 will if muxed with NanDub.

Since I am not highly experienced with divx I didn't mention it as I feared the answer would be "That's why it is called OggMux, you dummy!" :) .

Anyway, great job on OggMux. It has a clean, simple and powerful interface that is highly intuitive and easy to use (assuming you press the add button) ;)

[Sven Bent]

Anyway, great job on OggMux. It has a clean, simple and powerful interface that is highly intuitive and easy to use (assuming you press the add button)

Yes this tools ROCKS.
-simple
-usefull
-logical

That why i using it. Much nicer especial to show friends how to mux, whos doesn't have the big knowledge or can't remmeber long "guides"
FU + oggmux would be my favorit combo for newbies wanting to "backup" dvd's and drop the old flask way.
Muxing with nandub seem to "scary" them the most when they se such an advanced/complex program.

[tangent]

Fatboy and Spahm are not pre-echo clips but impulse clips. Although they are both solved using the same methods (short block switching), they are actually different things to detect. The classic pre-echo clip is Castanets, and Vorbis performs very well with it compared to MP3 at the same bitrates.

[Koepi]

Erm. Ok, if this is about _me_ fixing the mp3-crap with OGG, then please tell me _how_ I could do this?

Tell me what you do in graphedit to make a _separate_ video and mp3 stream get muxed with tobias' ogg mutliplexer dsf and _works_ afterwoods.

If you know of a method that can achieve this, I'm willing to implement it.
If the solution is first muxing that to an avi and afterwoods....
NO. I won't do such a crap. I prefer to delete the mp3 filetype from the list of supported sound types if it doesn't work at all. Same with aac, if you don't come up with a solution that makes it work _without_ doing some crappy avi-muxing before i can't support it.

As I wrote some XX times, OggMux does _exactly_ the same that you do with GraphEdit. So the solution must be in GraphEdit itself, else I can't support it.

Regards,
Koepi

[Sven Bent]

Originally posted by tangent
Fatboy and Spahm are not pre-echo clips but impulse clips. Although they are both solved using the same methods (short block switching), they are actually different things to detect. The classic pre-echo clip is Castanets, and Vorbis performs very well with it compared to MP3 at the same bitrates.

just show how little I know form the technical/program point of view of audio encoding as i stated. :)
but thanx for clearign that up. i can now use the correct therm of the ogg.vorbis "weak spot"


Erm. Ok, if this is about _me_ fixing the mp3-crap with OGG, then please tell me _how_ I could do this?

Don't get me wrong. I have NO idea if this problems lies in oggmux or in the ogg contain or the mp3 DSF or what ever. As i said
"3: Might have koepi to fix this issues.. if it at all is an issue with oggmux and not the ogg container itself"

I only said that IF the problems was related to oggmux and not ogg itself, THEN i would apreciate that you might look into it.
if it is a problem with the oog container or the mp3 DSF there is not much you can do besides using the same crapy way nandub does (but it works).

My post was NEVER meant to insult you or in anyway blame you for not doing a god job. Because i really consider oggmux of a great tool.
but as i read you message i understand that the probelms lies outside oggmux.
and that was the only answer i needed...

[Mikele]

@Koepi:
Maybe you should remove .mp3 and .ac3 from the supported audio formats list, they don't work anyway and then nobody will think it's a OggMux bug.

[Neo Neko]

I don't think that should be done. OggMux for what it does is quite simple and should incurr no bugs. It is simply a more friendly directshow interface to the current filters in use. There is not that much that could or should go wrong with OggMux itself. The thing is that Directshow can not propperly handle VBR mp3 period. And AAC support is limited ATM. But I don't think that koepi should remove them. I think maybey he should add a read me explaining this so that people would not get the misconception that it is his program.

[Mikele]

People don't read readmes :)
But what is wrong with AC3, why doesn't it work? I've read that the filters are locked, but this has nothing to do with AC3 decoding filters, it's Ogg Multiplexer that does not accept these files.

[MaTTeR]

I have to agree with Neo Neko. The options should remain in the tool as we are currently testing these formats.

I would love to know what this phrase means about locked AC3 filters. I've seen it a few times but nobody is explaining it. Can someone please clarify? :)

In testing AC3 with Ogg, as near as I can tell in GraphEdit, the AC3 file is always dumping PCM data. I haven't figured a work around out yet but I don't give up easily:D