Interest in surround encoder improvements? Options

[NullC]

Jean-marc has been doing a bunch of work improving the encoder for surround streams (e.g. on sensible rate allocation and such). I haven't seen much discussion of surround here— anyone here care about surround opus at all? Should I put up some test binaries?

[Dynamic]

I suspect it will become more wanted as video codecs start to permit opus audio in future, which I'd imagine some will.

I guess there are a few folks here who rip surround audio from music DVDs that might wish to encode to Opus, but few surround setups are portable enough to absolutely require low bitrate audio unless it's for streaming.

I think a lot of us would support moves to make Opus ready for primetime in future video standards or would have enough curiosity to test them, but personally I don't have a surround setup at present so I can't help.

[IgorC]

Google will launch its new video format VP9.

[CoRoNe]

Video codecs don't permit audio codecs, containers do. Although experimental, you can already mux opus in matroska.

[C.R.Helmrich]

Sorry, totally off-topic, but the relevant people are reading this topic: is the Daala project related in any way to the VP9 undertaking? Are the R&D forces combined, or are these two separate approaches by different entities (Google vs. Mozilla/Xiph/...)?

Chris

[bawjaws]

Sorry, totally off-topic, but the relevant people are reading this topic: is the Daala project related in any way to the VP9 undertaking? Are the R&D forces combined, or are these two separate approaches by different entities (Google vs. Mozilla/Xiph/...)?

Chris

I'm not affiliated with either project in any way, but it did seem like there was some (at least potential) mingling of the two in regards to the IETF's internet video codec project which, as one option had simply adopting VP8 or 9 as it stands, or of combining the best ideas from various parties to create a new codec. I don't think any final decision was made and it's all gone a bit quiet from the outside. (Opus of course itself being the result of the IETF internet audio codec project, and the combination of Skype's silk and Xiph's CELT codecs.) But apart from that potential mingling they're fairly distinct projects as far as I can see.

http://www.ietf.org/mail-archive/web/video...t/maillist.html

This is (in my non-lawyerly opinion) complicated by the patent situation. Google seems to have mostly bought themselves out of trouble, but mostly in a limited "if you use the tech for VP8/9 you're okay" way, but that doesn't mean you can fork or independantly reinvent that tech without further negotiations.

Related to some other things said above, experimental WebM with VP9 video and Opus audio support is apparently already in Chrome nightlies since late last year. They just announced that they plan to have VP9 finalised and "code checked into Chromium / Youtube" by the middle of June.

https://groups.google.com/a/webmproject.org...uss/UzoX7owhwB0

This post has been edited by bawjaws: May 9 2013, 14:54

[IgorC]

Sorry for offtopic but since we talk about VP9, if somebody is interested here some results for H.264/HEVC/VP9 http://forum.doom9.org/showthread.php?p=1614397#post1614397

It's a bit old and unfair result since VP9 hasn't a code freeze while HEVC does.
Anyway imo Google developers will need to work hard to get VP9 on par with HEVC.

This post has been edited by IgorC: May 9 2013, 16:51

[xiphmont]

Sorry, totally off-topic, but the relevant people are reading this topic: is the Daala project related in any way to the VP9 undertaking? Are the R&D forces combined, or are these two separate approaches by different entities (Google vs. Mozilla/Xiph/...)?

Chris

Technologically, only a little bit. We (the Xiph guys) and Google are pretty much in the same room on a regular basis. In a sense these are rival efforts with different development strategies. VPx is taking an incremental, tried-and-true approach to wringing more performance out of traditional DCT-based codecs (and there's plenty more wringing to do, especially given further strides in available CPU). Daala is trying something substantially new that's higher risk and also a potentially higher reward. It may or may not succeed, we'll find out.

This post has been edited by xiphmont: May 18 2013, 03:27

[wswartzendruber]

When you say that JMV is improving surround encoding, do you mean that he's altering the bitrates for the various frames inside the multistream packets? It seems like this could better distribute available bits.

This post has been edited by wswartzendruber: Jun 7 2013, 07:13

[BFG]

(random speculation) This conversation makes me wonder just how "portable" a 22.2 signal container would be with current technology. (/random speculation)