what I found more is some frequencies (such as 16.2k, 16.3k) may contribute
more errors in the lower octave band than other frequencies (such as 16k)
do.


>I have the anti-aliasing filter before downsampling.
>
>any idea what's wrong?
>
>>the problem is when there is a high frequency (such as 16k) sine wave
>get
>>into my octave filters. the octave 16k band is fine while there are
>>unwanted high output from other octave bands (such as 31Hz band 63Hz
>>band,)
>>I am expecting -70dB from these lower bands.
>>
>>
>>>On Tue, 20 Oct 2009 09:05:41 -0500
>>>"boozy" <boozy@sina.com> wrote:
>>>
>>>> I use decimation filter plus IIR filters to do octave filters
>>>> (20Hz-20kHz). Sample frequency is 48K. I found the filter shape
>>>> distortion at high frequency band (16k). I thought this might be
>>>> related with the sample frequency, so I increase the sample
>>>> frequency, but nothing is improved. what's wrong?
>>>> 
>>>> 
>>>
>>>Simple, you screwed something up.  If you were to post your Octave
>>>code, and be a bit clearer as to what you're trying to accomplish,
>>>someone might even be able to take a stab at what.
>>>
>>>-- 
>>>Rob Gaddi, Highland Technology
>>>Email address is currently out of order
>>>
>>
>

I have the anti-aliasing filter before downsampling.

any idea what's wrong?

>the problem is when there is a high frequency (such as 16k) sine wave
get
>into my octave filters. the octave 16k band is fine while there are
>unwanted high output from other octave bands (such as 31Hz band 63Hz
>band,)
>I am expecting -70dB from these lower bands.
>
>
>>On Tue, 20 Oct 2009 09:05:41 -0500
>>"boozy" <boozy@sina.com> wrote:
>>
>>> I use decimation filter plus IIR filters to do octave filters
>>> (20Hz-20kHz). Sample frequency is 48K. I found the filter shape
>>> distortion at high frequency band (16k). I thought this might be
>>> related with the sample frequency, so I increase the sample
>>> frequency, but nothing is improved. what's wrong?
>>> 
>>> 
>>
>>Simple, you screwed something up.  If you were to post your Octave
>>code, and be a bit clearer as to what you're trying to accomplish,
>>someone might even be able to take a stab at what.
>>
>>-- 
>>Rob Gaddi, Highland Technology
>>Email address is currently out of order
>>
>

the problem is when there is a high frequency (such as 16k) sine wave get
into my octave filters. the octave 16k band is fine while there are
unwanted high output from other octave bands (such as 31Hz band 63Hz
band,)
I am expecting -70dB from these lower bands.


>On Tue, 20 Oct 2009 09:05:41 -0500
>"boozy" <boozy@sina.com> wrote:
>
>> I use decimation filter plus IIR filters to do octave filters
>> (20Hz-20kHz). Sample frequency is 48K. I found the filter shape
>> distortion at high frequency band (16k). I thought this might be
>> related with the sample frequency, so I increase the sample
>> frequency, but nothing is improved. what's wrong?
>> 
>> 
>
>Simple, you screwed something up.  If you were to post your Octave
>code, and be a bit clearer as to what you're trying to accomplish,
>someone might even be able to take a stab at what.
>
>-- 
>Rob Gaddi, Highland Technology
>Email address is currently out of order
>

On Tue, 20 Oct 2009 09:05:41 -0500
"boozy" <boozy@sina.com> wrote:

> I use decimation filter plus IIR filters to do octave filters
> (20Hz-20kHz). Sample frequency is 48K. I found the filter shape
> distortion at high frequency band (16k). I thought this might be
> related with the sample frequency, so I increase the sample
> frequency, but nothing is improved. what's wrong?
> 
> 

Simple, you screwed something up.  If you were to post your Octave
code, and be a bit clearer as to what you're trying to accomplish,
someone might even be able to take a stab at what.

-- 
Rob Gaddi, Highland Technology
Email address is currently out of order

I use decimation filter plus IIR filters to do octave filters (20Hz-20kHz).
Sample frequency is 48K. I found the filter shape distortion at high
frequency band (16k). I thought this might be related with the sample
frequency, so I increase the sample frequency, but nothing is improved.
what's wrong?