On Sun, Aug 23, 2009 at 05:44:55PM +0100, Dan Mills wrote: > The VU is a slow response meter (300ms integration time IIRC)intended to > (badly) track perceived volume,
The VU specs are based on what was possible 50 or more years ago. A typical VU would just be a diode bridge and a series resistor feeding a moving coil meter, and no active electronics (you had to drive them using a separate line amplifier to avoid distortion caused by the diode bridge). That means it measures the average of the absolute value (and not RMS), and the dynamics are defined by a second order filter acting on the rectified value. For a sine signal corresponding to 0 VU the meter should reach 99% in 300ms, and overshoot by between 1 and 1.5% before returning to the correct value. > The differences each time you play are why we leave the thick end of > 20db of headroom between 0 VU and 0dbFS, you should (in a production > environment) never be going anywhere near 0dbFS (there is no need for it > in the age of 20+ bit ADC noise floors). VUs are not very useful today. On one side, the digital domain does not overload as gracefully as magnetic tape, so you should be conservative and use e.g. -20dB digital = 0 VU. On the other hand some of today's commercial music is so heavily compressed that a VU calibrated like that and used on a final mix would go off scale well before there is any digital overload. In that case you'd want a different calibration, even less than the historical one which was around 9 dB below 'peak', with peak being a somewhat softer limit than it is today. So if you still want to use a real VU it almost has to have variable calibration, which is a bad idea as it doesn't allow you to form any stable mental picture of the relation between loudness and the meter indication. Combining a true RMS and a digital peak meter in one display without any gain difference between the two is IMHO the best way to go, and it is exactly what e.g. the K-meter does. Ciao, -- FA Io lo dico sempre: l'Italia รจ troppo stretta e lunga. _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/mailman/listinfo/linux-audio-dev
