Ac3 encode question

epirb wrote on 2/28/2004, 5:29 PM

When I render to .ac3 for output to DVDA , I usaully start with the stereo dvd template. I then go to the custom settings and change in the "preprocessing " tab change both dynamic range compression setting to >none<.
My question is on the "audio service" tab should I leave the dialog normalization at -27 or set it to something else like -31 or -1?
the audio is basicly narration over music bed.
also is what is the peak level I should make sure the narration is at in the original track?

Comments

Spot|DSE wrote on 2/28/2004, 5:57 PM

Unfortunately, due to the way AC3 encoders work, you can't hear a proximate 'post encode' signal. The metadata included in an AC3 stream uses the dialog normalization as the decode instruction set, so if it's set wrong, everything else will be weird. -31 is a common setting, in the AC3 forum you'll find a bunch of discussion on this. Anyway, for a stereo mix, this isn't critical like it is for a surround mix that will be summed/downmixed. So...our house standard for stereo is -31, range to none, dialog being loudest track set to approximately -10 dB of overall mix, masters set to 0, using an Ultramaximizer plugin set to -03. That encodes quite nicely. Unfortunately, Dolby won't open this up much....so it's doubtful you'll ever see a post-encode monitoring. I've asked a few people for a plugin that would approximate the behavior of compression, but everyone says that unless Dolby will provide more info, they don't see how to estimate it.

farss wrote on 2/28/2004, 6:04 PM

What I've taken to doing is using the RMS value from SF as the dial norm setting. I'm only following others advice and so far it seems to work OK.
It's quite remarkable though how different the values can be from program to program. None of it was stuff I'd mixed myself, some of it already had heavy compression so the reading was around -7dB, other material had no compression and came in at around -24dB.

Spot|DSE wrote on 2/28/2004, 6:11 PM

Interestingly enough, because of some recent research, I was told by one of the ATSC guys that "vegas is one of the extreme few tools that does this right."
Sound Forge is a good gauge. It's a great tool, isn't it?

epirb wrote on 2/28/2004, 6:16 PM

thanks guys,
farss the rms reading your getting is that importing one soundfile consisting of all audiotracks? if I understand you right, then thats what you set the normalization setting in the ac3 customs tab? or the setting for the norm switch in vegas properties ?
i guess what I'm not sure is I have two audio tracks obviously the narr and the music. so I would need the rms value for the mix of the two correct?