Existence of any TRUE 5.1 ch compression formats?

[SkaDood1]

Ive been extensively reading alot of posts about multichannel development lately. Ive been searching for and testing many of the solutions on here.

Is there ANY 6-ch fully compressed(coupled and all, meaning seperate channels may use main channels to build off and compress greatly) codecs in existence out there at all? or is vorbis the only one making an attempt at it?

Thanks for any input:-)

[DSPguru]

there's also mpeg2 multichannel, aac multichannel, mp3 multichannel.
check out the audio faq, q13, for info about mp2mc.

[Antimon]

Would not ac3 qualify?

See chanal coupling and compresion is nice and all, but if it doensn't interface with the majority of hardware out there properly, it doesn;t really matter.

Personally i NEED spdif out...if the format can;t deliver that which none can cus of hardware limitations curently, i may as well stick with pro loogic and i think a lot of people are in my boat plus ac3 really isn't all that big..it's a much higher compresion ratio with quality then mp2 is.

[MaTTeR]

Originally posted by Antimon
[B]
Personally i NEED spdif out
Whats wrong with using multi-channel analog? If you have a decent sound card such as Terratec 6Fire, M Audio Dio, etc then you might find that the analogs will give better output then the digital SPDIF. I learned a long time ago that SPDIF/digital isn't always the best as I had assumed:)

True Multi-Channel Coupling is being worked on already for Vorbis. In fact once this is finished then you will really see how important it is and watch your file sizes drop dramtically. Channel Coupling isn't designed or meant to interface with any hardware, it will simply just compress common components between channels much better. So again this really wont have anything to do with hardware, in the time being we can just use analog ports which again, are not bad and sometimes even better providing your not using a subpar card.

For the time being, I'll stick with AC3 and continue to push the Vorbis devs to get multi-channel coupling implemented. I've done very little testing with AAC but my first impression is that it really isn't meant for prime time yet, maybe it will prove me wrong in further tests:D

[LigH]

Dolby Digital (AC3) is usually the first choice when planning to encode multi-channel sources with reasonable quality and effective channel coupling; the only drawback is that the AC-3 codec is directly comparable to MPEG1 Layer 3 in its efficiency regarding the relation of compression factor and audible quality therefore it requires a relatively high bitrate to achieve good quality with many channels - to give you some kind of thumb rule: 448 kbps for 5.1 source is comparable in quality to a 160-192 kbps stereo MP3.

448 kbps for audio is indeed a bit too high for a 1-CD movie rip; therefore it is obvious that the hunt for better compression technologies is just beginning - and the results at Ogg Vorbis are already very promising.

MPEG2-MC is not a real multi-channel solution: due to the requirement of being compatible to MPEG1 audio (which supports stereo at max), MPEG2-MC cannot use a complete channel coupling, the extended channels have a limited relation to the stereo channels - therefore it requires more bandwidth (up to 912 kbps) for 5.1 audio. MPEG4-AAC instead will be able to use multi-channel-coupling - provided that encoders are supporting it.

dts is also capable of encoding multi-channel sources, but with an even lower compression factor - fortunately, 5.1 sources are usually compressed to the same bitrate as a stereo CD wave audio would have. (I'm not sure how much the channels are related / coupled in dts...)

[Antimon]

@Matter My amp only has 4 point suround inputs *front rear* i dont have the center/sub input. So the only way to get 5.1 is through spdif, unless iw anna spend another $300 on an amp and speakers.....

I fidn it difficult to belive that 6 analog inputs will sound better then a spdif input, i can only asume the decoder in most amps is better then the dsp on a soundcard for 5.1 I'd more easily believe they are equal, or the soundcard being better.
I have an sblive 5.1 x-gamer and a cambridge soundworks desktop theature 5.1 *first gen befer the upgrade*

[Prosper]

I should hope that AC3 already does channel coupling, else even a 448k AC3 only leaves 89.6kbps per channel.

[int 21h]

Originally posted by LigH
Dolby Digital (AC3) is usually the first choice when planning to encode multi-channel sources with reasonable quality and effective channel coupling; the only drawback is that the AC-3 codec is directly comparable to MPEG1 Layer 3 in its efficiency regarding the relation of compression factor and audible quality therefore it requires a relatively high bitrate to achieve good quality with many channels - to give you some kind of thumb rule: 448 kbps for 5.1 source is comparable in quality to a 160-192 kbps stereo MP3.

448 kbps for audio is indeed a bit too high for a 1-CD movie rip; therefore it is obvious that the hunt for better compression technologies is just beginning - and the results at Ogg Vorbis are already very promising.


Using the new MCF-CD advancements, 448 kbps for audio barely makes a difference on a 1 CD rip in terms of video bitrate.

I disagree with your conclusion about Ogg. Currently, if you take into account the LFE (which you really shouldn't because it is bandwidth limited anyways), AC3 runs at around 64kbps per channel. (Its VBR though, so it varies). It encodes with some very very sophisticated psychoacoustic models, allowing for channels to be combined and seperated, and using a global bit allocation technique, it allows for any channel to borrow bits from anywhere they aren't being used. There is absolutely no way, in its current state, Ogg, even with channel coupling, could beat an AC3 in terms of quality and compression ratio. You're talking about encoding each channel at less than 64 kbps. To even reap a significant gain, you would need to encode somewhere near 48kbps or 56kbps at most. Honestly, the very small bandwidth gain is not nearly enough incentive to go through an entire transcoding process (converting from lossy to lossy codecs is never a good idea if it can be avoided).


MPEG2-MC is not a real multi-channel solution: due to the requirement of being compatible to MPEG1 audio (which supports stereo at max), MPEG2-MC cannot use a complete channel coupling, the extended channels have a limited relation to the stereo channels - therefore it requires more bandwidth (up to 912 kbps) for 5.1 audio. MPEG4-AAC instead will be able to use multi-channel-coupling - provided that encoders are supporting it.

dts is also capable of encoding multi-channel sources, but with an even lower compression factor - fortunately, 5.1 sources are usually compressed to the same bitrate as a stereo CD wave audio would have. (I'm not sure how much the channels are related / coupled in dts...)

DTS is usually encoded in loseless mode. It has an effective bandwidth of 1509 kbps or a half rate of 754 kbps. The only advantages DTS has over Dolby is that you can encode in a loseless mode, and that it handles all 6 channels discretely. Obviously, for compression's sake, we can't handle all our channels discretely in a 'backup' environment.

[LigH]

@int 21h:

It is of course possible that I misunderstood some details - therefore, thank you for correcting obviously wrong statements.

But your central statement about Ogg is something I might have to agree in the current state of development, but I absolutely expect changes in near future:


There is absolutely no way, in its current state, Ogg, even with channel coupling, could beat an AC3 in terms of quality and compression ratio.


The reason: It is wrong to linearly divide the overall bitrate by the amount of (full) channels; as far as I read, AC-3 uses a technology like joint-stereo in MP3, due to the channel coupling of mid-side stereo encoding, it is possible to require less than twice the bitrate of a mono audio for the stereo audio with a perceptually equal quality. And a higher amount of channels profits even more from this technology: a 448 kbps 5.1 channel AC-3 can sound as good as a 192 kbps stereo AC-3 but requires even less than 5/2 the bitrate (LFE not even added). As a thumb rule, I read in a book, you can say that the factor for n channels to multiply to the base bitrate of an equally well sounding mono audio is more like the squareroot of n instead of n.

In one of my former posts, I wrote about an analogue point of view about mid-side encoding: You can imagine that the sound has some kind of common "kernel", like the trunk of a tree (this is the "mid" part), and several channel-specific parts, like the branches of a tree (the "side" part of stereo is just the difference between left and right, the "side" parts of multi-channel audio may need to be a bit more complex).

Due to the fact that Ogg Vorbis can provide an almost transparent sound for stereo material at 80-96 kbps for most listeners, where the best MP3 encoders still need at least 160 kbps (and AC-3 is very similar to MP3 regarding the encoding technology), I'm quite sure that Vorbis will have a good chance to achieve a similar quality at a lower bitrate for multi-channel audio with a similar bitrate relation. Maybe not in the current state of development, but hopefully in a not so far future.

[int 21h]

Originally posted on the Dolby Website
f. Carrier precombination
In general, the average bit demand of multiple channels appears to be roughly proportional to the square root of the number of channels. If it takes 128K bps to code a single channel, on average 5.1 channels require 128*sqrt(5.1) = 289 Kbps, comfortably within the 320 Kbps minimum datarate used by AC-3. This implies most multichannel signals can be properly coded via just the flexibility of the global bit allocation technique. For high demand signals, selective precombination of high frequency carrier components is used to provide further coding gain.

To my ears, Ogg at 80-96kbps is terrible. However, I am one that in most cases prefers a good DTS track for a movie I really love. When you are looking for the best quality, only discrete channel handling and loseless compression will do (usually DVD has the bandwidth to cover this), however, I truly believe that when the newer CD format takes off for MP4 in mode 2, the amount of AC3 usage in DivX backups will sky rocket. Avoiding transcoding is of utmost importance to me in most cases, and as media costs plummet, and bandwidth increases (in small leaps and bounds in some places, like the US, (consumer environments, not places like Universities or workplaces), its already huge in other places) I really think the push will not be towards how much we can fit in how little, it will be how accurately can we reproduce the original.

Anyways, look how long it took LAME to get a good joint stereo routine, now multiply the complexity of that by 6, and it will give you a good idea of how long it will take for Ogg to do the same. (especially if they use an approach similar to Dolby, et al. and use a truly joint bit reservoir).

In any event, I'm not here to tell anyone what to use, I was just simply relating the technical facts of Dolby and DTS, and what I personally think. To each his own, whatever sounds good to you, etc. This is not a socialist or communist movement by any means :p

Take it easy.

[LigH]

No problem - everyone is allowed to tell us his opinion! :)

But I hope you can see that 1-CD movies need all the bitrate they can get for the video - the bitrate requirement of a 5.1 Dolby Digital sound it too much for a 1-CD movie. Therefore many people have to be satisfied with a Dolby Surround audio today; a real multi-channel audio is currently not recommendable for such a low movie size - this will require at least a 2-CD rip to give the video enough bitrate.

Those who really want top quality shall buy the DVD. But those who are happy with a 1-CD rip usually just need a satisfying audio quality.

Fun and success! :cool:

[int 21h]

Originally posted by LigH
No problem - everyone is allowed to tell us his opinion! :)

But I hope you can see that 1-CD movies need all the bitrate they can get for the video - the bitrate requirement of a 5.1 Dolby Digital sound it too much for a 1-CD movie. Therefore many people have to be satisfied with a Dolby Surround audio today; a real multi-channel audio is currently not recommendable for such a low movie size - this will require at least a 2-CD rip to give the video enough bitrate.

Those who really want top quality shall buy the DVD. But those who are happy with a 1-CD rip usually just need a satisfying audio quality.

Fun and success! :cool:

I believe that the 100mb increase in media size obtainable from burning in true mode 2 (i.e. MCF-CD), produces nearly enough room for adding a 5.1 surround sound track instead of an mp3 track. It definitely produces enough room to always include the 2 channel Dolby track (if available) and on a 2 cd encode, 5.1 Dolby Track.

[Antimon]

I find 128 mp3's most listionable and near cd....*shrugs*

i alwuys find 160+ mp3's to be a waste, espeiclaly in stereo mode.

The only tiem 160 soudns better then 128 is when decodign in pro logic *to my ears*

and even then a good 128 lame sounds better then an average 160 in say radium.