On Sat, 9 Jun 2001 andy at theasis.com wrote: > > Now decibels are not anything as far as Psychophysics -- whatever that is! > > OK, what I was referring to is a (possibly manufactured) recollection of a > correspondence between 1 dB and the minimum difference detectible by human > audio perceptual apparatus. JND is measured in dB, that's all. > > ;) -- decibels are simply one tenth of a bel, which is a unit of relative > > change. They are very handy for electrical, acoustical, and a couple of > > other log scale measurements, > > Keep in mind that there is in fact always a reference against which to > compare any sound that hits an eardrum or a microphone membrane: the > sound pressure level caused by air molecules bouncing around on the > membrane. See below. You don't have to use them to measure sound. > Anyway, the above correspondence I mentioned does exist, but it may be > true that it just happened to work out that 1dB over background is pretty > close to the minimum detectible change in sound pressure level. OTOH, I'm > probably just wrong. I'll have to go check on that one. Not necessarily. It's just that JND has to specify the conditions. You can blow those conditions right out of the water with a strategic small change in environment. In any case, decibels just simply are never *defined* in terms of JND, JND is measured in dB. There are lot's of books to look at for confirmation. > > but you could just as easily say that Joe who > > makes $20k/year earns 6dB$ less than Sue who makes $40k/year. > > I think I disagree. First, I think it has to refer to power. Nope. Can refer to whatever you want. Power is the most common use, however. > Really, though, as soon as you get outside of power (into Voltage or > Sound Pressure), you're dealing with additional translations. Sound pressure is power. Power dB is 10log(ratio.) Voltage is 20log(ratio), which comes from Ohm's law directly. Any dB measurement of signals in electronics is by convention voltage (20log). > Also I think you want to say s/6/3/ dB, since log10(2) = .301, meaning a > doubling of _power_ (or $) is 3.01 ~ 3 dB. True. > Um... Remember that power is proportional to (voltage)^2, so using 6dB > would be appropriate if you were talking about voltage instead of power -- > a doubling of voltage is a 6dB increase or a power ratio of 4. You have a lot of the right pieces, but you're trying to force a voltage situation into power. I was. :) > > You have to specify the units, which in this case are understood to be > > dBFS (decibels w.r.t. Full Scale -- whatever $MAXVALUE is for your > > wordlength.) > > Ahh, ok. Except the part of being understood. As mentioned above, my > inclination is not to reference Full Scale by default. Yeah, but you're used to inverse squared power losses in location bat recording. Audio inside electric boxes should make you think 20log ratios, and digital should make you think dBFS as one of the major idioms. Analog signals should make you think voltage as dBV (0dB = 1V RMS into 600 ohms), dBu(pure voltage, ref level 1V, IIRC), or the most popular, dBm (*voltage* referenced to 1mW of power through 600 ohms.) > > Does that make it make more sense? > > That last part does clarify it quite a bit. Also you did force me to think > hard enough to organize the ratio thing a bit more. But the power > vs. voltage thing still leaves me wondering how this software is operating > on the sampled signal in the .wav. Correctly! I'll bring a book the next time we get together. It's a little bit confusing until you sort out all the different dB scales. -- "To misattribute a quote is unforgivable." --Anonymous