Hello there,
The synthesis QMF equation which is used in SBR can be converted accordingly to an FFT format and can be implemented which can reduce lot of machine cycles.
If anybody has worked on it, then i request you to share little knowledge.. I am getting some problems with the equationsss
Full Version: Optimisation to ARM
This paper might help you understand:
"Fast decomposition of filterbanks for the state-of-the-art audio coding"
Shih-Way Huang; Tsung-Han Tsai; Liang-Gee Chen;
Signal Processing Letters, IEEE
Volume 12, Issue 10, Oct. 2005 Page(s):693 - 696
Abstract: This paper derives fast decomposition for the quadrature
mirror filterbanks (QMFs) of the low power spectral band
replication (SBR) tools in the MPEG high efficiency advanced
audio coding (HE AAC) decoder. In contrast with the standard
method where computation-intensive matrix operations are employed
in the QMF, the proposed method decomposes the matrix
operations into conventional discrete cosine transform of type
II and III (DCT-II and DCT-III) and simple permutations for
easy implementation. The computational complexity can be also
reduced effectively by using fast algorithms for DCT.
Same authors published similar paper for www.AES.org (preprint number: 6336, convention: 118 (May 2005))
"Fast decomposition of filterbanks for the state-of-the-art audio coding"
Shih-Way Huang; Tsung-Han Tsai; Liang-Gee Chen;
Signal Processing Letters, IEEE
Volume 12, Issue 10, Oct. 2005 Page(s):693 - 696
Abstract: This paper derives fast decomposition for the quadrature
mirror filterbanks (QMFs) of the low power spectral band
replication (SBR) tools in the MPEG high efficiency advanced
audio coding (HE AAC) decoder. In contrast with the standard
method where computation-intensive matrix operations are employed
in the QMF, the proposed method decomposes the matrix
operations into conventional discrete cosine transform of type
II and III (DCT-II and DCT-III) and simple permutations for
easy implementation. The computational complexity can be also
reduced effectively by using fast algorithms for DCT.
Same authors published similar paper for www.AES.org (preprint number: 6336, convention: 118 (May 2005))
Does that mean the current decoders avalaible on the Palm OS (and PPC?) are not ARM optimized?
Vorbis use a lot less CPU to decode than AAC (HE-AAC use even more)
(in AeroPlayer/RealPlayer/PocketTunes)
Vorbis use a lot less CPU to decode than AAC (HE-AAC use even more)
(in AeroPlayer/RealPlayer/PocketTunes)
QUOTE(Daijoubu @ Dec 5 2005, 01:55 PM)
Does that mean the current decoders avalaible on the Palm OS (and PPC?) are not ARM optimized?
not necessarily.. It may be optimised to some other target. But the SBR QMF part takes the majority of the CPU time in AAC decoder than the other modules because of the matrix complexity involved in it.
Vorbis use a lot less CPU to decode than AAC (HE-AAC use even more)
Yes, true.
(in AeroPlayer/RealPlayer/PocketTunes)
not necessarily.. It may be optimised to some other target. But the SBR QMF part takes the majority of the CPU time in AAC decoder than the other modules because of the matrix complexity involved in it.
Vorbis use a lot less CPU to decode than AAC (HE-AAC use even more)
Yes, true.
(in AeroPlayer/RealPlayer/PocketTunes)
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.