FPGA Central

adamchapman wrote:
> On Feb 6, 9:45 pm, adamchapman <adamchapman1...@hotmail.co.uk> wrote:
>> Hello,
>>
>> I was reading this paper (http://www.kecl.ntt.co.jp/icl/signal/
>> nakatani/papers/icslp2002f0.pdf) and thought his Power spectra looks
>> ridiculously high.
>>
>> using 10*log10(S) for the log power of spectrum S, S would need to be
>> about 10^9 to get that high dB range.
>>
>> Or is there some standard that I'm unaware of?
>>
>> In my matlab code, 10*log10(S./min(S)) gives a similar sized spectrum
>> as long as min(S) is not zero of course.
>>
>> Any clarification would really be appreciated,
>>
>> Adam
>
> Sorry I should have noted that Im looking at the second row of plots
> in figure 2 of the linked paper.

Seems harmless enough, if somewhat unorthodox (not normalized). The
plots are shown from 100dB down to 20dB, a range of 80dB which is well
within the range of even 16bit samples; by convention audio treats
digital peak as 0dB, and "normalizes" integer samples to the +-1.0
range, so that one would expect to draw such a plot from 0dB to -80dB -
the pictures themselves will be unchanged.

Richard Dobson