@valex

[DSPguru]

reading this (http://forum.doom9.org/showthread.php?s=&threadid=27718), this (http://forum.doom9.org/showthread.php?s=&threadid=37852) and this (http://forum.doom9.org/showthread.php?s=&threadid=39137) makes me think -
how about extending your ac3filter project into a more general one ?


main ideas are :
a. adding AC3 encoding. (mostly of 640kbps AC3 (http://sourceforge.net/projects/ac3encode) for spdif playback of multichannel audio).
b. adding MPx decoding. (based on MAD (http://www.mars.org/home/rob/proj/mpeg/)).
c. adding AAC decoding (based on FAAD2 (http://www.audiocoding.com)).

[rjamorim]

Great idea! :D

I would gladly switch OdioDekoda for such a tool.

[Valex]

Yes, I thought about it. And I think one more format needed: DTS (most requested). I think it will be not too hard (except the searching for information).

I keep watching to ac3encoder thread and will take a part in this project.

[bergi]

Is it possible to make also an ACM Codec? I think then we would need the VD AC3 hacks.

[ChristianHJW]

I may be wrong here, but i guess an AC3 ACM codec was a pretty bad idea ( correct me if i'm wrong ), even with AC3 being generally CBR, not ABR/VBR . Not only from a technical approach, but also seen from a licensing perspective, especially when distributing binaries from sourceforge.net . Dolby could sue sourceforge for large license payments in first instance, as they are really the distributing entity ( and EVERY court would confirm this ) and sourceforge will aproach valex then to get their money back, as every project admin is signing that he, and he alone, is responsible for the content and its compliance with international legislation. Please bare in mind Dolby is NOT Fraunhofer in this respect, ask rjamorim about the nice letters he got from them about the AAC encoding binaries on his rarewares site.

But i loves DSPguru's approach about the real time AC3 encoding to export Vorbis 5.1 via SPDIF to external surround receivers.

A good question to investigate how the licensing situation would look like here, as the encoded AC3 files are not stored at all, but AC3 was more or less only used as a 'transportation layer' for the Vorbis 5.1 track ....

Dolby would certainly argue that this doesnt matter at all and their patents were used to encode, so its subject to licensing payment. A counter argumentation could be to appeal to the market domination of Dolby in this respect ( no surround receivers in the market would even understand 5.1 PCM via SPDIF as an alternative, although this was technically easily possible ), so the opensource community ( and amybe even Xiph, if they would be interested ) could argue that its necessary to encode to AC3 as transport layer contemporarily to reach a workable status of economical competition. There is a good chance the courts would follow this approach, especially when looking at Dolby's monopolism.


A much better and eaiser approach, however, would probably be to handle ac3filter as an 'educational project' and not distribute binaries from the sf.net project page, but leave compiling and distributing binaries to people living in countries where Dolby's patents ar enot registered :) .... anybody :D ??

[Valex]

2ChristianHJW

I may be wrong here, but i guess an AC3 ACM codec was a pretty bad idea ( correct me if i'm wrong ), even with AC3 being generally CBR, not ABR/VBR

It does not matter CBR or VBR AC3 is. And I think that it is possible to make decoder-only ACM codec (it will get an error for encoding). It will be useful for much of VFW applications (especially VD).

But i loves DSPguru's approach about the real time AC3 encoding to export Vorbis 5.1 via SPDIF to external surround receivers.

Dolby may argue, that as it uses DirectShow it can be easily used to store encoded stream. So it is not just a transport layer. (?)

A much better and eaiser approach, however, would probably be to handle ac3filter as an 'educational project' and not distribute binaries from the sf.net project page, but leave compiling and distributing binaries to people living in countries where Dolby's patents ar enot registered .... anybody ??

I tried to do this (in Distribution section):
This application may solely be used for demonstration and educational purposes. Any other use may be prohibited by law in some coutries. The author has no liability regarding this application whatsoever. This application may be distributed freely unless prohibited by law.

Is it enough? And how distributing binaries affects it?

I think DTS decoder may also have licensing problems. (?)

[rjamorim]

Originally posted by Valex
Yes, I thought about it. And I think one more format needed: DTS (most requested). I think it will be not too hard (except the searching for information).

HAHA. The only possibility of DTS support is reverse engineering. It's probably among the most closed formats ever. You won't find sources, libraries, specifications or anything else. Well, maybe you can somehow find leaked libraries from Cyberlink or Intervideo - but that's all.

[rjamorim]

Originally posted by ChristianHJW
Not only from a technical approach, but also seen from a licensing perspective, especially when distributing binaries from sourceforge.net . Dolby could sue sourceforge for large license payments in first instance, as they are really the distributing entity ( and EVERY court would confirm this ) and sourceforge will aproach valex then to get their money back, as every project admin is signing that he, and he alone, is responsible for the content and its compliance with international legislation.

Well, you can take the same approach of Lame, FAAC, MPEG4ip, XviD, etc: Don't distribute binaries!

Please bare in mind Dolby is NOT Fraunhofer in this respect, ask rjamorim about the nice letters he got from them about the AAC encoding binaries on his rarewares site.

Eheh:
http://www.inf.ufpr.br/~rja00/Dolby.txt

A much better and eaiser approach, however, would probably be to handle ac3filter as an 'educational project' and not distribute binaries from the sf.net project page, but leave compiling and distributing binaries to people living in countries where Dolby's patents are not registered :) .... anybody :D ??

Hehe. I can host them here, no problem. :)

I tried to do this (in Distribution section):
This application may solely be used for demonstration and educational purposes. Any other use may be prohibited by law in some coutries. The author has no liability regarding this application whatsoever. This application may be distributed freely unless prohibited by law.

Is it enough? And how distributing binaries affects it?

Well, it's enough to keep you in a slightly safe position, but, of course, Dolby can still start a legal battle. They have the lawyers and the money for that. It's always a risk.

Regards;

Roberto.

[ChristianHJW]

Originally posted by Valex 2ChristianHJW :I may be wrong here, but i guess an AC3 ACM codec was a pretty bad idea ( correct me if i'm wrong ), even with AC3 being generally CBR, not ABR/VBR
It does not matter CBR or VBR AC3 is. And I think that it is possible to make decoder-only ACM codec (it will get an error for encoding). It will be useful for much of VFW applications (especially VD).

Hmmm .... Nandub/VirtualdubMod already can read AC3 files and mux into AVI, and for AC3 processing i highly recommend BeSweet or HeadAC3he in any case .. what exactly was the benefit from having an AC3.acm to be able to decode AC3 in AVI in normal Virtualdub ?

But i loves DSPguru's approach about the real time AC3 encoding to export Vorbis 5.1 via SPDIF to external surround receivers.
Dolby may argue, that as it uses DirectShow it can be easily used to store encoded stream. So it is not just a transport layer. (?)

Fully correct ! On the other hand, i guess its very well feasible to make a Dshow filter that way that it blocks itself when writing to a file writer instead of SPDIF out. Also, if i understand DSPguru correctly than this AC3 encoder should use modified sources to be as fast as possible ( we are talking about realtime Vorbis decoding and AC3 encoding, for 6 channels ! ), so it would probably be necessary to simplify the PSY model ( or however it is called in Dolby terminology ) to save precious CPU cycles. This wouldnt hurt at all if you are outputting in the highest possible bitrate for AC3 standard ( i thought 556 kbps is highest DSPguru ? ), with 640 kbps you have more than enough bytes to make it sound really nice, even with a more simplistic PSY model !
Now, how does this change the view from Dolby about the filter ? Well, if output bitrate of the filter is fixed to 640 kbps its easy to argument that writing such a huge AC3 file to any file is pointless :D !!

A much better and easier approach, however, would probably be to handle ac3filter as an 'educational project' and not distribute binaries from the sf.net project page, but leave compiling and distributing binaries to people living in countries where Dolby's patents ar enot registered .... anybody ??
I tried to do this (in Distribution section):
This application may solely be used for demonstration and educational purposes. Any other use may be prohibited by law in some coutries. The author has no liability regarding this application whatsoever. This application may be distributed freely unless prohibited by law.
Is it enough? And how distributing binaries affects it?

I am not sure here, sorry, so i better shut up :( . But honestly speaking, i dont think the disclaimer alone would save your ass. I am pretty sure that Doom9 could tell us, he must have investigated this closely i guess ....

I think DTS decoder may also have licensing problems. (?)

You can bet on that ... not sure how Lucas would react about DTS being reverse engineered .. but i somehow doubt he would be pleased about it ???

[DSPguru]

Originally posted by rjamorim
HAHA. The only possibility of DTS support is reverse engineering. It's probably among the most closed formats ever. You won't find sources, libraries, specifications or anything else. Well, maybe you can somehow find leaked libraries from Cyberlink or Intervideo - but that's all. not exactly ;) (http://forum.doom9.org/showthread.php?s=&threadid=34958)
Originally posted by ChristianHJW
i thought 556 kbps is highest DSPguru ?Table 5.18 (http://www.atsc.org/standards/a_52a.pdf).

[rjamorim]

Originally posted by DSPguru
not exactly ;) (http://forum.doom9.org/showthread.php?s=&threadid=34958)

Humm... OK.

But are there decoding specs in that document (I don't really feel like reading it), or just frame headers specs?

Anyway, I guess you should prioritize faadlib2 and madlib support. The libraries are ready, so adding them to the filter will hopefully be a snap. :)

[bergi]

If you make an AC3, MP3, DTS decoding filter, i tink you have to split your source into modules, so why not make also AC3 encoding modul and an AC3 dummy encoding modul. If you want to share binary just compile it with the dummy encoder.