How does it work audio surrond encoder ?

[tiziano]

Dear all

I'm sorry for the question but I'm not able to find a clear answer.
Please, anyone can tell me how surround audio encoder works ?
I mean what kind of operations matrix does ?
having: L+C+R+LS+LR+LFE in what manner matrix sum and subtract each channel to L and R ?
I thank you very much for your help

[star882]

Just add some of the LFE and C to the L and R.

[tiziano]

Originally Posted by star882

Just add some of the LFE and C to the L and R.

Dear Star, the problem is that a matrix is not just an adding ... so I have not clue how the channels are together .. what is the scope of my information ?
I would like to build a matrix my self, I would like to design one ... because I don't find any Integrated Circuit that does the same ... I tried to looking for it without any result So I was thinking how to build a matrix with operational amplifiers ... but I have not clue what kind of mathematic operations are needed in order to matrix L & C & R & LS & RS & LFE just to L & R in a way that each person with a Dolby Dec, can decode it ... perhaps I'm too stupid but if I now the math relationship I can build something .. thank you very much

[andy_c]

Hi,

There's a bunch of white papers linked from this page of the Dolby web site. Some of them have block diagrams of the encoders and decoders, but the information may not be enough to design one yourself.

Edit: Here's a link to the technical library, which allows you to search for the PDF files with the documents.

[tiziano]

Originally Posted by andy_c

Hi,

There's a bunch of white papers linked from this page of the Dolby web site. Some of them have block diagrams of the encoders and decoders, but the information may not be enough to design one yourself.

Edit: Here's a link to the technical library, which allows you to search for the PDF files with the documents.

Hi Andy thank you very much but unfortunately the second link contains only product specs ... and first link just principles nothing helpful to design a circuit ... :-(

[Key]

Here is an intro.

http://en.wikipedia.org/wiki/Matrix_decoder

But keep in mind that info is incomplete and in some cases truncated. Most of the equations use irrational numbers so for instance when you see a .7 it really should be √ 1/2.

Personally I don't think encoding works.

[tiziano]

Originally Posted by Key

Here is an intro.

http://en.wikipedia.org/wiki/Matrix_decoder

But keep in mind that info is incomplete and in some cases truncated. Most of the equations use irrational numbers so for instance when you see a .7 it really should be √ 1/2.

Personally I don't think encoding works.

Fantastic .. so If I have understood correctly:

To encode I have to add to channel L

• squareroot of the half of the amplitude relative tu the channel C ( 0.707 * C )
• the squareroot of 2/3rd of amplitude about rear Left ( 0.81 RL ) obtainable through resistor divider and shifted +90°
• the squareroot of 1/3rd of amplitude about rear Right ( 0.577 RR ) obtainable through resistor divider shifted +90°

The same thing over the channel R but all shifting are delayed by 90°
Is that correct ?

If yes: The LFE channel ? wher is it ?

And: how to shift then 90° ( constant shift ) ? By a delay line I suppose ... but I have not clue what analogue delay can do this ...

[andy_c]

Originally Posted by tiziano

Hi Andy thank you very much but unfortunately the second link contains only product specs ...

On the left, click on "technologies" rather than "products". Then choose for example "Dolby ProLogic II". Doing so led me to this page, which has links to, among other items, this PDF file.

[tiziano]

Originally Posted by andy_c

On the left, click on "technologies" rather than "products". Then choose for example "Dolby ProLogic II". Doing so led me to this page, which has links to, among other items, this PDF file.

Thank you Andy ... indeed the principle is contained ... so ... I have to correct my previous answer ... +J and -J = 90° shifting but in one case ( +J ) added, in the other case ( -J ) subtracted ... anyway this diagram report four sources L-C-R-S and not 5 or better L-C-R-LR-LS I'm curious to know the LFE in what manner is added then ... As delay line I think a "all pass filter" is applicable ... what do you think ?

[Key]

Originally Posted by tiziano

And: how to shift then 90° ( constant shift ) ? By a delay line I suppose ... but I have not clue what analogue delay can do this ...

Here is the weakest link imo. I don't think a linear 90 phase shift is a realistic thing. But of course being a hard headed human I tried to make a couple haha. In analog I believe the way to do it would be using all-pass networks to approximate 90 across the entire bandwidth. Digital I believe you would want to use impulse responses FIR and IIR. None of these methods that I have heard yield perfect results imo.

The LFE imo would be on the decoding side. Some of the phase shift can be made more trivial by the addition of a derived crossover before the decoder - passing the bass as stereo but decoding everything above the crossover point.