On 25 Mai, 18:04, Kennedy McEwen <(EMail Removed)> wrote:
> In article <(EMail Removed)>, Floyd L. Davidson
> <(EMail Removed)> writes
>
> >Kennedy McEwen <(EMail Removed)> wrote:
> >>In article <(EMail Removed)>, Floyd L. Davidson
> >><(EMail Removed)> writes
> >>>Kennedy McEwen <(EMail Removed)> wrote:
> >>>>In article <(EMail Removed)>, Floyd L. Davidson
> >>>><(EMail Removed)> writes
> >>>>>Kennedy McEwen <(EMail Removed)> wrote:
> >>>>>>In article <(EMail Removed)>, Floyd L. Davidson
> >>>>>><(EMail Removed)> writes
> >>>>>>>ransley <(EMail Removed)> wrote:
> >>>>>>>>On May 22, 5:28*am, Marc Wossner <(EMail Removed)> wrote:
>
> >>>>>>>>> SNR = 20 log (Signal RMS / Noise RMS)
>
> >>>>>>>dB has no particular attachment to sound, any more than
> >>>>>>>it does to light. *It's a ratio of two numbers, as shown
> >>>>>>>in the formula above. *The numbers represent signal
> >>>>>>>power, regardless of what the signal actually is.
> >>>>> * * * * ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>
> >>>Read what I wrote very carefully Kennedy; it should be
> >>>very apparent that right from the start I know exactly
> >>>what you are referring to.
>
> >>I have, and I think you will realise that, from the start I have
> >>considered what you wrote to be ambiguous at best or indeed, if your
> >>words are interpreted in a logical manner but not necessarily the manner
> >>you meant, completely wrong.
>
> >>"Its a ratio of two numbers" followed by "the numbers" (ie. the SAME
> >>numbers) "represent signal power, regardless..."
>
> >>That is completely wrong. *"The numbers" do NOT represent signal power,
> >>they are voltages. *The ratio (signal / noise) is a VOLTAGE ratio.
>
> >A decibel is a power ratio, by definition.
>
> The DECIBEL is, but NOT the ratio of the numbers you referred to.
>
> >>The ratio (signal^2 / noise^2) IS a power ratio,
>
> >End of discussion.
>
> >...
>
> >>>It is a *power* ratio. *That is true regardless of what
> >>>the two numbers actually represent.
>
> >>No it isn't!
>
> >It is a power ratio by definition.
>
> Only if you follow the definition. *If you don't, such as by using
> numbers without regard to their meaning, exactly as you suggested, then
> you are not following the definition and therefore end up with something
> that is NOT a power ratio.
> 
> Kennedy
> Yes, Socrates himself is particularly missed;
> A lovely little thinker, but a bugger when he's ****ed.
> Python Philosophers * * * * (replace 'nospam' with 'kennedym' when replying)
As this highly theoretical discussion is not so easy to follow for me,
IŽd like to bring it back to something that I can work with: If I
chose the numerator of snr to be the range available in the image
file 256 for 8bit or 4096 for 12bit  and the denominator to be
the standard deviation as the equivalent to RMS noise, is the form SNR
= 20 log (Signal RMS / Noise RMS) correct or should it be SNR = 10 log
(Signal RMS / Noise RMS)?
Best regards!
Marc Wossner
