On Fri, Jul 23, 2010 at 02:20:02PM +0200, Arnold Krille wrote: > But if two people talking is power because their talking is uncorrelated > (even > if they speak about the same thing:), then making a PA louder (twice as loud > as before) would be an amplitude thing as it is correlated. No?
Assuming you can find some gain factor that your test listeners would agree on to call 'double', then that will correspond to some amplitude ratio R and a power ratio R^2. It doesn't matter wich one you use. The important thing here is that your test is measuring the subjective loudness effect of playing back the *same* sound at different levels, as opposed to increasing the level by adding other sounds. I guess that if you ask people to adjust the levels of two similar sounds until one seems 'double as loud' as the other, what they do is choose some ratio that they feel comfortable with calling 'double' in everyday parlance. I doubt very much if such a choice would be consistent with their choice for other (bigger) loudness ratios. For example if you find that for the your listeners 'twice as loud' means +10dB, and then you'd play two sounds with a difference of 40dB, would they experience that as '16 times as loud' ? I doubt if they would even be able say if it is 8 or 32 times. There is another consequence of calling some dB difference 'twice as loud'. If the result is consistent (in the above sense), it means that our perception of loudness is *not* logarithmic as it is often described to be: if it were, a change of a fixed number of dB would correspond to a constant subjective *difference*, and not to a constant ratio such as 2:1. For example, the classical +10dB for double loudness means that L = c * P^0.3, L = loudness, P sound power, and c some constant. This is a power law, not a logarithmic one. If 'twice as loud' would correspond to +6dB, that would actually mean that loudness is a *linear* function of sound pressure... All this makes me think that the whole concept of 'twice as loud' is meaningless, and where it has been used it is a bad name for something else. Ciao, -- FA There are three of them, and Alleline. _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/listinfo/linux-audio-dev
