Help - Search - Members - Calendar
Full Version: subband filtering in MP2
Hydrogenaudio Forums > Lossy Audio Compression > Other Lossy Codecs
kangaroo
hello everyone,
I have a doubt in Mp2 filterbank module ...
filterbank in Mp2 uses band pass filtering , i mean the samples are divided into 32 subbands ...
since i have never gone thru this codec ...i have a doubt regarding wether there is any FFT involved in implementing filterbank module for Mp2.
And in Mp3 its Hybrid filterbank that means Mp2 filterbank + MDCT
if we use FFT in calculating the Mp2 filterbank then we use FFT for MDCT also ...
is it true ...
it will be helpful if some one clears my doubt ...

Thanks in advance ......

bye......
kwwong
QUOTE(kangaroo @ Feb 13 2006, 01:01 AM)
hello everyone,
I have a doubt in Mp2 filterbank module ...
filterbank in Mp2 uses band pass filtering , i mean the samples are divided into 32 subbands ...
since i have never gone thru this codec ...i have a doubt regarding wether there is any FFT involved in implementing filterbank module for Mp2.
And in Mp3 its Hybrid filterbank that means Mp2 filterbank + MDCT
if we use FFT in calculating the Mp2 filterbank then we use FFT for MDCT also ...
is it true ...
it will be helpful if some one clears my doubt ...

Thanks in advance ......

bye......
*



Well, you can actually implement the 32 subband filters using a single FFT. I am not very sure of the "recommended implementation" as indicated in the ISO specifications. (It has been a long time since I last look at the MP2 specs)

It is also true that the MDCT can be implemented using FFTs, but I doubted that this is used here since the length of the MDCTs are very short, which mean that there is not much computational savings implementing them using FFTs. wink.gif
Woodinville
The spec for the coder does not make it clear how to do this filterbank efficiently.

There is a good AES paper out there from a few years ago (conference preprint) on how to do that. Sadly I don't recall when.
kwwong
QUOTE(Woodinville @ Feb 16 2006, 06:12 PM)
The spec for the coder does not make it clear how to do this filterbank efficiently.

There is a good AES paper out there from a few years ago (conference preprint) on how to do that. Sadly I don't recall when.
*



I wondered, is the 32 subband filters a Perfect Reconstruction filterbank? I didn't do any studies on them. unsure.gif
Woodinville
QUOTE(kwwong @ Feb 17 2006, 07:16 PM)
QUOTE(Woodinville @ Feb 16 2006, 06:12 PM)
The spec for the coder does not make it clear how to do this filterbank efficiently.

There is a good AES paper out there from a few years ago (conference preprint) on how to do that. Sadly I don't recall when.
*



I wondered, is the 32 subband filters a Perfect Reconstruction filterbank? I didn't do any studies on them. unsure.gif
*



It's not, but it's within this close || of a system that is, and that can be done with a few twiddles.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2008 Invision Power Services, Inc.