Raimund Specht wrote:
>>>Does anyone know if CoolEdit Pro's spectrogram would add
> spurious harmonics?
> Hi John,
> I still have not seen any spurious harmonics in CoolEdit. In fact, I
> can not remember of any other major program having such problems.
> However, theoretically it would be possible.
I have seen such harmonics in CoolEdit sonograms put up by others. As
well as other errors.
> A simple test would be to generate a synthetic sine signal, that
> does not have any harmonics. When your spectrogram software does not
> show harmonics on that signal, your software would be ok.
For sine test signals only. Much better to work out the things your
sonogram program does on the stuff you record.
> Filtering would not be a reliable test for that. The reason is, that
> there is no ideal low-pass filter, that would remove the signals
> from the stop-band (the frequency range above the cutoff frequency)
> completely to 100 percent. The stop-band attenuation of common
> filters is limited to something like 60 or 90 dB (or even higher),
> depending on the type of filter. So, larger signals may remain in
> the filtered file (at much lower amplitudes of course). These
> remaining signals (or harmonics) visible on a spectrogram were not
> produced by the specrogram algorithm. They are really present in the
> sound file.
Funny, my sonogram programs show differences as small as a few dB quite
clearly. Showing a change of 60-90 dB is extremely easy, in fact a
sonogram program that cannot show such a change would be absolutely
useless. A error produced by a sonogram calculation will not filter out
at all, leaving it unchanged.
I consider any sonogram program that does not have some sort of
calibrated dB scale to be a antique. Most color sonograms can be
calibrated, or already are.