AAC | ADPCM | Convolution | DAFx | FFT | IIR | Mixer | MP3 | MPEG | MPEG-4

Hi, rsv,
Thank you for your kind reply. can you tell me where I can find the hanning window equation?
ISO11172-3 didnot offer it.

regards
ZiP

-----????-----
???: Prasad R S V [mailto:rsv@rsv@...]
????: 2000?11?29? 13:01
???: ZiP HE
??: audiodsp@audi...
??: Re: [audiodsp] a MPEG2 question
	
 Hi,
   
   Generally,the remaining 128 samples are ignored. But the ignored
samples are not consecutive.The first 64 samples ( 0 to 63) and last 64
samples ( from 1088 t0 1151 ) are ignored and the central 1024 samples are
used for FFT.
  It is not mandatory that we should use 1024 samples for FFT..You can use
1152 or 2304 ( previous plus present frame samples) samples for analysis
purpose. It will not effect the bit stream which specified in the
standard, but computational complexity will increase.Thta's why people
suggest 1024 as the magic number for FFT in MPEG -1 layer 2 &3.

rgds
rsv 
	On Tue, 28 Nov 2000, ZiP HE wrote:

> Dear all,
> I am a newcomer in DSP, and now working on MPEG-2 Audio Encoder. In ISO11172-3 encoding
algorithms(Layer II and Layer III), before FFT transfer, there are 1152 samples collected, but
only 1024 points FFT are employed, what shall we do with the remaining 128 samples, just
ingnored or shall overlapped by Module Reduction N?
> Before FFT an Hanning window is firstly applied.
> 
> Best Regards!
> 
> ZiP
> 28/11/00
> 
> -------------------------- eGroups Sponsor -------------------------~-~>
> eLerts
> It's Easy. It's Fun. Best of All, it's Free!
> http://click.egroups.com/1/9699/1/_/414121/_/975431838/
> ---------------------------------------------------------------------_->
> 
> To Join:  audiodsp-subscribe@audi...
> 
> To Post:  audiodsp@audi...
> 
> To Leave: audiodsp-unsubscribe@audi...
> 
> Archives: http://www.egroups.com/group/audiodsp
> 
> Other DSP-Related Groups: http://www.dsprelated.com
> 
>
	

(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )

ZiP HE wrote:

> I am a newcomer in DSP, and now working on MPEG-2 Audio Encoder. In ISO11172-3
> encoding algorithms(Layer II and Layer III), before FFT transfer, there are 1152
> samples collected, but only 1024 points FFT are employed, what shall we do with the
> remaining 128 samples, just ingnored or shall overlapped by Module Reduction N?
> Before FFT an Hanning window is firstly applied.

Your particular application may be different, but commonly, zeroed
samples are added at the end to complete an FFT window.  I.e., add
(1024 - 128) zeroed samples.

> Thank you for your kind reply. can you tell me where I can find the hanning window
> equation? ISO11172-3 did not offer it.

The Hamming window is a special case of the Hann window.  Given an
array of samples, samples[s_size], a window array, window[w_size],
a pointer i to each array element, the Hamming window is:

for (i=0; i<w_size; i++) {
   window[i] = samples[i] *
              (1.0 - (0.54 + 0.46 * cos(2.0 * PI * i / w_size));
}

PI = 3.141592653589793116
2 * PI = 6.28318530717958465692
cos is the cosine function.

A Hann window is where you choose any different values in the
above for 0.54 and 0.46, but where they still sum to 1.

It is faster when using the window repeatedly to initially build
an array of just the Hamming values.  It is a good idea to graph
the resulting array to confirm that it looks like a bell shaped
curve.

If you want to get the correct amplitude value from the FFT output
for each frequency, those amplitudes will need to be mulitplied by:
the area under a rectangular window divided by the area under a
Hamming window, approximately (1.0 / 0.46).

Regards,

Neil Nelson
	

(You need to be a member of audiodsp -- send a blank email to audiodsp-subscribe@yahoogroups.com )