Dan, I do agree, like most of you, that peak normalization is not a good volume balancer along the whole catalog. Nevertheless it's a simple strategy in avoiding clipping. I think that many Rivendell users kept the default -12 dB peak normalization, without tuning this value according to the loudness of the song. I am far from being an expert on this field, but I think that once the risk of clipping has been avoided, a good 'modus operandi' would be leveling the catalog volume according the loudness, and in particular with the Ebu128 standard. Considering I have already imported some thousands of tracks, all with the default -12 norm, I was thinking of using the gain value in the cut table to pimp a bit those tracks with high dynamics which suffered from the peak normalization more than others. Can you think it is a good strategy? If I am not wrong only values in the range of +/- 6dB are accepted in that field.
Alessio 2014-02-16 21:10 GMT+01:00 Fred Gleason <[email protected]>: > On Feb 16, 2014, at 14:26 27, Dan Mills <[email protected]> wrote: > >> Looks like I might be once again working in broadcast in the not too >> distant future (Mostly SDI video toys rather then audio this time), so >> I thought I had better re up, life happened as it so often does.... > > Hey Dan, welcome back! > > >> Fred, would you take a patch to make the normalization use the R.128 >> loudness estimation method rather then the rather simple minded >> measure it uses at present, it should make for much more even loudness >> across cuts? The spec does not look hard to implement, but it would >> really benefit from the meter having a Loudness unit mode as well as >> VU. >> This would among other things make for much better leveling between >> compressed music and speech cuts. >> >> I think I can cook up a measurement code that can be included in the >> source so removing the library dependencies issue. > > It sounds good. Does this mean that we'd wind up with compliant meters as > well? That'd be sweet. > > >> I figure an extra entry in the cuts table with the computed LU value, >> and probably a target level in the groups table so that different >> things can be trimmed to slightly different average loudness levels >> (Naughty, but someone is going to want to go there). > > Is there a reason we shouldn't keep the current normalization target scheme > (a system-wide default, with the ability to override that on a one-off basis)? > > Cheers! > > > |-------------------------------------------------------------------------| > | Frederick F. Gleason, Jr. | Chief Developer | > | | Paravel Systems | > |-------------------------------------------------------------------------| > | A room without books is like a body without a soul. | > | -- Cicero | > |-------------------------------------------------------------------------| > > _______________________________________________ > Rivendell-dev mailing list > [email protected] > http://caspian.paravelsystems.com/mailman/listinfo/rivendell-dev _______________________________________________ Rivendell-dev mailing list [email protected] http://caspian.paravelsystems.com/mailman/listinfo/rivendell-dev
