Excerpts from Aurelien's message of 2011-05-12 10:07:48 +0200: > On Tue, May 10, 2011 at 11:30:24AM -0400, Paul Davis wrote : > > On Tue, May 10, 2011 at 11:14 AM, Joe Hartley <jh@xxxxxxxxxxxx> wrote: > > > On Tue, 10 May 2011 10:51:26 -0400 > > > Paul Davis <paul@xxxxxxxxxxxxxxxxxxxxx> wrote: > > >> the data path during export is EXACTLY the same as during normal > > >> playback except for the conversion to an integer format. since > > >> metering occurs before this, the data the metering displays is > > >> IDENTICAL to the data that would be seen during normal playback. > > > > > > It would seem to me that if the meters are checking their inputs > > > n times per second, then if an export is occurring and is running say > > > 3 times faster than real time, there is the possibility that a peak > > > could happen at a moment that the input isn't being checked, and thus > > > the peak levels could be reported differently during export than in > > > real time. ÂI wouldn't expect to see a huge difference in the reported > > > values, but I can see how this would happen. > > > > > > Does this make any sense, or have I misunderstood how the meters > > > get their values? > > > > well, the numerical values should match because the actual "peak" > > values are based on all data since the last meter update. but its > > definitely true that the "graphical" part of the meters could be > > different for precisely the reason you mention on the other hand, if > > there *is* a difference, its mostly likely to be an error with > > metering and not with any difference in the data generated for export. > > OK, not sure to have all understood, but, by the way, when speaking > about vu-meters, I obviously mean "numerical values". It's true that I > just trusted those values, and never compared them to the data generated > for export. > > So, just to be sure, I just have to ignore values of meters (including > numerical ones?) when exporting, and refering to the ones appearing when > listening to my mixes before exporting? > > By the way, I would have to retry (and I have no time for this today), > but I'm almost sure to have got different numerical values from one > "realtime listen" to another. I especially remember a session with a > zone in which I sometimes get huge saturation (+1dB or more), and the > time right after got -0.8dB. It was a very low dynamic region (metal > music), and I had to get it the loudest I could, and it was quite a > pain. You should try sndfile-info <exported_file>, it's a quick way to get the real peak. Of course it would be nice if ardours metering was correct, even on export. _______________________________________________ Linux-audio-user mailing list Linux-audio-user@xxxxxxxxxxxxxxxxxxxx http://lists.linuxaudio.org/listinfo/linux-audio-user