Re: Why is more power required for BASS?

On Sat, 02 Jun 2007 08:00:07 -0700, Mark <makolber@xxxxxxxxx> wrote:



Beware of FFT - that shows energy, not power.


Perhaps I should clarify the term FFT. I am talking about an FFT
software based spectrum analyzer. These are typically calibrated in
dB indicating relative voltage and (assuming a fixed load resistance)
relative power. For example, a steady 1V RMS 1 kHz tone may display
as steady 0 dB. A 2 V RMS tone as +6 dB indicating twice the voltage
and 4x the power.

But this is power not energy. To display energy, the display would
have to integrate over time and with a steady input tone the display
would climb higher and higher the longer in time that the tone
persists indicating more and more energy has passed. The FFT's I have
seen do not do this but rather maintina a steady display if the tone
level is steady.

No, FFTs do exactly as you say. The longer a tone lasts the higher it
will be displayed. I am talking of course of tones of duration shorter
than the sampling period (sampling interval * length of FFT). Clearly
the display won't keep climbing because it gets reset when each new
sample set is started.

So they display relative power, not energy.

Sorry about the delay; I couldn't think of an easy way to explain this
in writing, so have a look at a little flash movie I just made using
Audition. It shows quite nicely why FFTs measure energy, not power
especially for music.

http://81.174.169.10/odds/fftoddity

d

--
Pearce Consulting
http://www.pearce.uk.com
.

Relevant Pages

  • Re: Why is more power required for BASS?
    ... Beware of FFT - that shows energy, not power. ... To display energy, the ...
    (rec.audio.pro)
  • Re: Why is more power required for BASS?
    ... Beware of FFT - that shows energy, not power. ... a steady 1V RMS 1 kHz tone may display ... A 2 V RMS tone as +6 dB indicating twice the voltage ...
    (rec.audio.pro)
  • Re: Human hearing instataneous dynamic rage?
    ... in making a visual display of your frequency vs time data. ... reason you're not seeing what you want is the length of the FFT. ... Another thing is the window used for the FFT. ... A dynamic range of 16 bits ...
    (comp.dsp)
  • Re: FFT for N Not a Power of 2
    ... remove the largest number of observations that are a power of 2, ... There are FFT algorithms that work for any N, including prime numbers, but ... with floor rising near the two tones. ... had to ensure that the length of the window and periods of each of the ...
    (sci.optics)
  • Re: FFT numeric range?
    ... netpbm only supports integers, although these ... of the range of values for FFT, ... Let's assume that you have an 'image' of floating point pixel values in the range and you want something to display in the range, then the following little conversion formula is your friend. ... I guess that FFTW will not object too loudly to input values in, but if it does, then scale the values to in a floating point array. ...
    (sci.image.processing)
Loading