On Fri, Feb 27, 2009 at 1:08 PM, robs <aquegg@yahoo.co.uk> wrote:

Stat appears to be broken for multi-channel audio: it works in MCHAN mode taking contiguous samples and performs ffts on them and deltas on them, neither of which is a valid thing to do when channels > 1.  The question is, how should stat handle this situation?  One possibility would be a table of stats with one column or row for each channel, and maybe an overall column for those stats where this makes sense.

Another question is: do we really need all these stats?  I'd have thought that peak, rms, and dc-offset were the main three of interest.

Yep, its all I use... The runtime VU meter works for most my other purposes.

Perhaps just document what user should be doing to get what they desire?  Using the mixer effect to grab one channel at a time or similar.

Whatever you think is best.