Sound Levels for DVD, question

[Vincent Oliver]

I have been setting my sound level to -2db for DVD production (mainly voice over work). Should I be using -6 or -12db?

Any advice or tips would be much appreciated.

[Rick Reineke]

As a finished product 'Video' DVD? Stereo or 5.1?
Technically... either way, as long as you do not exceed 0dBFS you should be OK. However I prefer I few dBs more headroom for conversion and unknown D/A converters.. of course the actual 'loudness' depends on how much it's compressed and other factors. There's usually no submission guidelines like there is for B'cast TV. For Instance: Ref. @ -20dBFS with program peaks not exceeding -10dBFS. There are LKFS meters available for the new 'CALM Act' ATSC standards as well.

If the VO track is a component to be used in editing.. it should delivered in a uncompressed PCM file format, with adequate levels which are not clipped or otherwise distorted.

[Tom Morrow]

There are lots of different standards, but to keep it simple I like to use the rule:

If this is going to be edited or processed down the line, leave 12db headroom (max peak -12dbFS).

If this is the final mastered product to be delivered to the end listener, then use as much of the range as you want to get the loudness to the ballpark of other competing products. It's better to hear a signal with only -1dbFS of headroom than the same signal compressed down to a "nice" -12dbFS level.

[Warren Kawamoto]

I set my peak levels at -12db max. You can tell this is in the ballpark by comparing it with any mass produced movie dvd.

Oh wait, I just read your post again. "Production dvd" means you have data on it meant for editing? If so, then don't exceed 0dB. The editor can pull it down in post if they need to.

[Vincent Oliver]

Thanks, I normally set the recording level to approx -12db and then Normalize the sound track to -2 for the final DVD master. It's the final Normalize figure I wanted to know about.

I produced a DVD recently of a musical for which I Normalized to -2db and added extra sound tracks (from 2 & 3 mikes). I produced 50 DVDs and picked one copy at random to check. Yep, the sound was dreadful I must have been overworking because I didn't check the combined levels before writing. I binned the 50 copies and reset the combined sound levels t0 -2db, no problem, sounds OK now.

I have made a substantial investment in sound gear this year, yet have never looked into recording techniques in any depth. So forgive me if I start asking a few seemingly basic questions over the next few weeks

[Steve House]

Be careful normalizing as it can bring everything up to a uniform level. Broad dynamic range with contrasts between loud and soft is part of the dramatic flow of a production or the musical expression of a performance. Applying a blanket 'correction' by normalizing to some arbitrarily chosen level can destroy that if you're not careful. Better to ride gain manually, with accurate metering, as you master the final mix.

[Vincent Oliver]

I was under the impression that Normalize just raised the peak levels to a set value and moved everything up in relation to that level, i.e. the soft passages would still remain soft relative to the peak level.

Correct me if I am wrong, but what you refer to is "Equalize" the level

[Steve House]

Consider you have two sections of dialog. One has an average level of -15 dbFS and the highest peak touches -10 dBFS. The other has an average level of -10dBFS with the highest peak touching -5dBFS. You normalize both with the max level set to -3 dBFS. The first clip gets 7dB of gain added, the second gets 2dB of gain added, bringing the max peak in each clip to -3 dBFS. But what has happend to the average levels? In the first clip, the average rises from -15 to -8. In the second, the average rises from -10 to -8. Both clips are at the same level and the dramatic distinction between them is lost. Is those clips were two different takes of the same scene and you're cutting between them that's what you want - their levels should match. But if clip one is a love scene at one point in the film and clip two is a fight scene at some other time in the movie, you defintely do not want the levels to be the same. Normalizing within a scene is fine. Normalizing the entire film usually is not.

[Vincent Oliver]

In Audition there is only the option to Normalize to a user defined value, there isn't a Peak or Loudness option

[Steve House]

Read how it is applied. Normalize looks at the level of the highest peak in the selection being processed and determines how much gain needs to be added to bring that peak to the desired level, then applies that gain uniformly over the entire selection in question. If your selection is the entire film, that means the same amount of gain is added over the whole piece from start to finish. You need to be sure that's really what you want to do. It could be that quiet passages are already just right and adding that much gain to them will make them too loud.

[Vincent Oliver]

Thanks Steve, I was aware that the entire selection would be moved up, putting the peaks at the chosen level.

This would surely just have the effect of raising the entire sound track level, leaving the dips relative to the highs. I guess the ideal solution would be to set the levels at the correct position. However, when capturing audio I understood that you have to leave some headroom. I do this by setting my recording level to -12 and then Normalize the sound to -2db, which in theory and in practice still leaves the loud and soft sections relative to each other. i.e. I do not get a sustained -2db throughout the production.

My first question was what level should I produce my DVDs at

[Vincent Oliver]

Maybe I am thinking about this in the wrong way Steve. What you say does make sense. I don't tend to make a global colour correction to my film, I do it section by section. Perhaps this is what I should also be doing with my audio.

Hmmm. I am thinking about it in a new way now.

Thanks

[Steve House]

Originally Posted by Vincent Oliver

Maybe I am thinking about this in the wrong way Steve. What you say does make sense. I don't tend to make a global colour correction to my film, I do it section by section. Perhaps this is what I should also be doing with my audio.

Hmmm. I am thinking about it in a new way now.

Thanks

Exactly! Normalize within a scene so that cuts from different takes have the same levels and intercut smoothly. But for the film as a whole, mix by ear on calibrated monitors and with properly calibrated meters (Dorrough loudness meters are a good bet). Put on a brick-wall limiter so you never accidently go over your target level, say -3dBFS or so.

[Vincent Oliver]

Thanks again Steve,

I have just ordered a book "Mastering Audio: The art and the science by Bob Katz", hope this will shed some light on audio techniques.

[Steve House]

Katz is a noted authority but do be aware that the tecniques and considerations for mastering audio for music CDs is a totally different art from film soundtrack mastering. Two highly recommended books are "Sound for Film and Television" and "Sound for Digital Video," both by Tomlinson Holman. You've surely encountered Holman's work, or his initials, before even though you might not recognise the name ... he's the "TH" in "THX Sound" and knows a thing or two about recording and mastering