Robin Bowes wrote: >> Which is why the power version uses a 20 log (x/y) and the >> normal definition uses 10 log(x/y) > > No, it's the other way round. > > Pressure levels use 20log(p/p0), power levels use 10log(w/w0)
Good catch, I get them switched all the time. And when I remember to switch it, I'm likely to switch from correct to wrong. >>> Anyeway, a 3dB change is not what I'd call a "barely perceptible change >>> in loudness". >> The usual definition of one dB is smallest reliably detected change in >> loudness. > > Er, I've not heard that definition before. It's certainly true, i.e. 1dB > is about the smallest reliably detectable change in loundness, but a dB > is not defined as that. Yeah, actually isn't it Bel that is defined, and dB is just a tenth of a bel? What I heard, could be wrong, is that a single dB is detectable, but it takes 3dB difference to have a 'significant' sample of users, say in a blind test, get it right. That's where the 'reliable detection' part comes in. Some folks can tell smaller, and at some frequencies, but when you want to make broad brush statements, you need about 3dB. Some 'modern' CDs are compressed to where there is barely 3dB difference anywhere in the tracks. -- Pat Farrell http://www.pfarrell.com/ _______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/lists/listinfo/audiophiles