Hi and sorry for the delay.
Artur Bodera wrote:
On Tue, Dec 1, 2009 at 12:50 AM, Vitor Sessak vitor1...@gmail.com wrote:
While I normally oppose making non-committed code more complex, I think
this feature is so often requested that it is worth the extra work in the
future. Stefano, Michael,
Author: alexc
Date: Sun Dec 6 23:48:39 2009
New Revision: 5499
Log:
10l: l_Temp = RATE*t_E'(L_E')
Modified:
aac-sbr/aacsbr.c
Modified: aac-sbr/aacsbr.c
==
--- aac-sbr/aacsbr.cSat Dec 5 22:21:28 2009
Author: alexc
Date: Sun Dec 6 23:48:41 2009
New Revision: 5500
Log:
Fix an off by one error. The phi matrix lacks column zero of the spec's \phi
and in the spec j is indexed from 1 = j 3.
Modified:
aac-sbr/aacsbr.c
Modified: aac-sbr/aacsbr.c
Author: alexc
Date: Sun Dec 6 23:48:42 2009
New Revision: 5501
Log:
Clean up the algebra in sbr_hf_inverse_filter using the principle that the main
diagonal of \phi as seen in the spec (i = j+1 in out matrix phi) is all real.
As a side effect many floating point overflows are avoided.
Author: alexc
Date: Sun Dec 6 23:48:44 2009
New Revision: 5502
Log:
Remove an unnecessary autocorrelation matrix position. phi[i+1][j] =
phi*[j+1][i] therefore phi[i][j] where i j+1 is unused.
Modified:
aac-sbr/aacsbr.c
Modified: aac-sbr/aacsbr.c
On Sun, Dec 06, 2009 at 11:48:43PM +0100, alexc wrote:
Log:
Clean up the algebra in sbr_hf_inverse_filter using the principle that the
main diagonal of \phi as seen in the spec (i = j+1 in out matrix phi) is all
real. As a side effect many floating point overflows are avoided.
Please
Author: alexc
Date: Mon Dec 7 02:27:31 2009
New Revision: 5503
Log:
Index the SBR Y matrix consistently.
Index Y[old/new][k][l][re/im] like all the other major SBR matrices.
This can be revisited for optimization purposes once the decoder works
properly.
Modified:
aac-sbr/aacsbr.c
Author: alexc
Revision: 5494
Property Name: svn:log
Action: modified
Property diff:
--- old property value
+++ new property value
@@ -1 +1,5 @@
-In envelope decoding i(k) is the replacement for k of the previous envelope
(it does not replace l). Ref 4.6.18.3.4 SBR envelope and noise floor
Author: alexc
Revision: 5500
Property Name: svn:log
Action: modified
Property diff:
--- old property value
+++ new property value
@@ -1 +1,4 @@
-Fix an off by one error. The phi matrix lacks column zero of the spec's \phi
and in the spec j is indexed from 1 = j 3.
\ No newline at end of file
Author: alexc
Revision: 5501
Property Name: svn:log
Action: modified
Property diff:
--- old property value
+++ new property value
@@ -1 +1,5 @@
-Clean up the algebra in sbr_hf_inverse_filter using the principle that the
main diagonal of \phi as seen in the spec (i = j+1 in out matrix phi) is all
Author: alexc
Revision: 5502
Property Name: svn:log
Action: modified
Property diff:
--- old property value
+++ new property value
@@ -1 +1,3 @@
-Remove an unnecessary autocorrelation matrix position. phi[i+1][j] =
phi*[j+1][i] therefore phi[i][j] where i j+1 is unused.
\ No newline at end of
11 matches
Mail list logo