[MPlayer-dev-eng] i need help - transformation matrix

[Arpi]

Hi,

> > >  63 22 44 90  4 81  6 31 86 58 36 11 68 39 73 53
> > >  69 57 18 88 34 71  1 23 46 94 54 27 75 50 30  7
> > >  70 92 20 74 10 37 85 13 56 41 87 65  2 59 24 47
> > >  79 93 29 89 52 15 26 95 40 17  9 45 60 76 62  0
> > >  64 43 66 83 12 16 32 21 72 14 35 28 61 80 78 48
> > >  77  5 82 67 84 38  8 42 19 51  3 91 33 49 25 55

> i have only 2 .rm files using 'sipr' audio, but they are same bitrate and
> so same buffersize.

hmm i've found one using 0x140 byte packets instead of 0x128.
after running the nibble-level 'cracker tool' :) i've got that it uses 40
nibbles (20 bytes) long sub-packets, and wonder: 0x140*2/40 == 16
and, it has 6 rows just like the 0x128 samples.
ie, it uses 16*6 sub-packets in teh transformation.
let's see the matrix it uses:

 63 22 44 90  4 81  6 31 86 58 36 11 68 39 73 53
 69 57 18 88 34 71  1 23 46 94 54 27 75 50 30  7
 70 92 20 74 10 37 85 13 56 41 87 65  2 59 24 47
 79 93 29 89 52 15 26 95 40 17  9 45 60 76 62  0
 64 43 66 83 12 16 32 21 72 14 35 28 61 80 78 48
 77  5 82 67 84 38  8 42 19 51  3 91 33 49 25 55

yes! it seems to be the same used by 0x128 variant.
so, even if they change buffer size, the number of sub-packets is constant
16 per row and they seems to stick to 6 rows per group.

at least for tehse 3 samples... i need much more samples :(

and probably this matrix is some hardcoded thing not generated by algo or
any logic, so we have to crack all possible variants and store it.
until now i have only this one, but the number of rows is a parameter in teh
header and it can be other than 6, at least the fileformat seems to allow
that.


A'rpi / Astral & ESP-team

--
Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu