Dears,
I wonder to know in those audio test equipment (e.g. CLIO, Audio Precision),
they have a function called Waterfall or CSD. It is for analyzing a speaker
time-frequency relation from an impulse. Do they use FFT or STFT to generate the
spectrum? Also, what windows do they use?
I tried to FFT an impulse with shifting square window but I got a lot of low
frequency noise after the first FFT. I wonder to know how the others do it.
Alvis
STFT or FFT Waterfall for loudspeaker?
Started by ●February 9, 2010
Reply by ●February 14, 20102010-02-14
No one know it?
Dears,
>
>I wonder to know in those audio test equipment (e.g. CLIO, Audio Precision), they have a function called Waterfall or CSD. It is for analyzing a speaker time-frequency relation from an impulse. Do they use FFT or STFT to generate the spectrum? Also, what windows do they use?
>
>I tried to FFT an impulse with shifting square window but I got a lot of low frequency noise after the first FFT. I wonder to know how the others do it.
>
>Alvis
Dears,
>
>I wonder to know in those audio test equipment (e.g. CLIO, Audio Precision), they have a function called Waterfall or CSD. It is for analyzing a speaker time-frequency relation from an impulse. Do they use FFT or STFT to generate the spectrum? Also, what windows do they use?
>
>I tried to FFT an impulse with shifting square window but I got a lot of low frequency noise after the first FFT. I wonder to know how the others do it.
>
>Alvis
Reply by ●February 14, 20102010-02-14
Elvis-
> I wonder to know in those audio test equipment (e.g.
> CLIO, Audio Precision), they have a function called
> Waterfall or
> CSD. It is for analyzing a speaker time-frequency
> relation from an impulse. Do they use FFT or STFT to
> generate the spectrum? Also, what windows do they use?
>
> I tried to FFT an impulse with shifting square window but
> I got a lot of low frequency noise after the first FFT. I
> wonder to know how the others do it.
Your description of what you tried seems fuzzy to me. What do you mean by "shifting"? How can your window be
"square" with amplitude and frequency both variable?
Suggest that you try STFFT with these params:
-approx 10 msec window size (e.g. if sampling rate
is 44.1 kHz, then window size is about 400-500
samples)
-1024 pt FFT size
-50% overlap (or higher, maybe up to 85%)
-Hanning window
Then post a plot so we can be sure your result is correct.
-Jeff
> I wonder to know in those audio test equipment (e.g.
> CLIO, Audio Precision), they have a function called
> Waterfall or
> CSD. It is for analyzing a speaker time-frequency
> relation from an impulse. Do they use FFT or STFT to
> generate the spectrum? Also, what windows do they use?
>
> I tried to FFT an impulse with shifting square window but
> I got a lot of low frequency noise after the first FFT. I
> wonder to know how the others do it.
Your description of what you tried seems fuzzy to me. What do you mean by "shifting"? How can your window be
"square" with amplitude and frequency both variable?
Suggest that you try STFFT with these params:
-approx 10 msec window size (e.g. if sampling rate
is 44.1 kHz, then window size is about 400-500
samples)
-1024 pt FFT size
-50% overlap (or higher, maybe up to 85%)
-Hanning window
Then post a plot so we can be sure your result is correct.
-Jeff
