On Sun, 24 Aug 2003, smoerk wrote:> there are different channel profiles for 5.1: > L R C LFE Ls Rs (DVD) > L R Ls Rs C LFE > L C R Ls Rs LFE (Film)I don't see why you need more than one of the above formats. It's not like you can't decide to put the LFE channel into "input 3" or "input 5" or whatever set of inputs your sending it to, depending on how that set of inputs is configured.> then we have ambisonics where we have 1st, 2nd and 3rd order (4, 9 and > 16 channels), but then you can also have 2 dimensional ambisonics > without heigt signal, which gives another 3 options, plus different UHJ > encoded variants, etc.Well, we already have the problem of being able to encode only a very limited number of independent tracks in this field, too. Currently the limit is 8, which does me no good when I want to FLAC-encode a multi-track "master" from my studio, which can easily be 48 tracks or more. One option would be to add a code for "get 16 bit number of tracks from end of header." (Don't laugh at 16 bits; I've got friends who have produced pieces with more than 256 tracks.) Given that, you might well just be able to add another code for "ambisonic, get 8-bit number of tracks from end of header," and then use the number of tracks to figure out the format. 1 W (kinda dumb way of doing monaural) 2 UHJ (stereo-compatable) 3 WXY (first order without height) 4 WXYZ (first order with height) 5 WXYUV (second order without height) 6 WXYUVW' (second order without height, mixed 1st/2nd order data) 9 WXYZRSTUV (second order with height) 10 WXYZRSTUVW' (second order with height, mixed 1st/2nd order data) 16 WXYZRSTUVKLMNOPQ (third order with height) (There can be no third order without height, since the coefficents of all of the third order equations have a height component. Note that I am just blindly using the equations from http://www.york.ac.uk/inst/mustech/3d_audio/secondor.html; one would want to check this with someone really in the know before standardizing on it.) Or one might want to tweak it to have every channel always in the same "slot," which is currently broken by the 2, 6 and 10 channel formats. cjs -- Curt Sampson <cjs@cynic.net> +81 90 7737 2974 http://www.NetBSD.org Don't you know, in this new Dark Age, we're all light. --XTC
Curt Sampson wrote:> On Sun, 24 Aug 2003, smoerk wrote: > > >>there are different channel profiles for 5.1: >>L R C LFE Ls Rs (DVD) >>L R Ls Rs C LFE >>L C R Ls Rs LFE (Film) > > > I don't see why you need more than one of the above formats. It's not > like you can't decide to put the LFE channel into "input 3" or "input 5" > or whatever set of inputs your sending it to, depending on how that set > of inputs is configured.i agree, but then flac needs a parameter to map the channels when converting from wav to flac or flac to wav. example: i'm working on a dvd production and my 5.1 files are L-R-C-LFE-Ls-Rs and FLAC's format is L-C-R-Ls-Rs-LFE. when encoding and decoding FLAC have to change the channel order of the file. but then we still have the possibility to use not all channels. think of L-C-R L-C-R-LFE L-R-Ls-Rs L-C-R-Surround L-R-Surround and more this is only for 5.1. i still think we need a more flexible approach than only defining 16 different channel mappings.>>then we have ambisonics where we have 1st, 2nd and 3rd order (4, 9 and >>16 channels), but then you can also have 2 dimensional ambisonics >>without heigt signal, which gives another 3 options, plus different UHJ >>encoded variants, etc. > > > Well, we already have the problem of being able to encode only a very > limited number of independent tracks in this field, too. Currently > the limit is 8, which does me no good when I want to FLAC-encode a > multi-track "master" from my studio, which can easily be 48 tracks or > more.FLAC can only encode 8 tracks? Why?> One option would be to add a code for "get 16 bit number of tracks from > end of header." (Don't laugh at 16 bits; I've got friends who have > produced pieces with more than 256 tracks.) > > Given that, you might well just be able to add another code for > "ambisonic, get 8-bit number of tracks from end of header," and then use > the number of tracks to figure out the format. > > 1 W (kinda dumb way of doing monaural) > 2 UHJ (stereo-compatable) > 3 WXY (first order without height) > 4 WXYZ (first order with height) > 5 WXYUV (second order without height) > 6 WXYUVW' (second order without height, mixed 1st/2nd order data) > 9 WXYZRSTUV (second order with height) > 10 WXYZRSTUVW' (second order with height, mixed 1st/2nd order data) > 16 WXYZRSTUVKLMNOPQ (third order with height) > > (There can be no third order without height, since the > coefficents of all of the third order equations have a height > component. Note that I am just blindly using the equations from > http://www.york.ac.uk/inst/mustech/3d_audio/secondor.html; one would > want to check this with someone really in the know before standardizing > on it.)only 5.1 encoded ambisonics is missing, i think it's called g-format. but i can ask in the sursound mailing list, if this options are sufficient.
Josh Coalson wrote:> --- smoerk <smoerk@gmx.de> wrote: > >>i didn't find anything about tagging flac files as surround files. i >>think there should some possibility to tell the player how to play a >>multi-channel file (how to map the different channels to the speakers >>or if there some decode is needed, like for ambisonic files). > > > it wouldn't be in the tags, it would be a specific channel > assignment in the frame header: > > http://flac.sourceforge.net/format.html#frame_header > > there are currently none defined for >2 channels but there could > be.i would write a proposal for this, but i don't think 4 bit is enough for all the different surround formats. for example in a ac3 encoder i can define any combination of left-right, center, surround and LFE channels. http://www.soundonsound.com/sos/nov01/images/surroundfig1dolbyblobs.l.gif there are different channel profiles for 5.1: L R C LFE Ls Rs (DVD) L R Ls Rs C LFE L C R Ls Rs LFE (Film) then we have ambisonics where we have 1st, 2nd and 3rd order (4, 9 and 16 channels), but then you can also have 2 dimensional ambisonics without heigt signal, which gives another 3 options, plus different UHJ encoded variants, etc. how can we squeeze this in 4 bits?