Walter Murch on Sound
 

http://www.transom.org/guests/review...iew.murch.html
The second article on sound mixing has some great examples from the movies he's worked on.

Some of that was quite interesting and some was kind
of hard to understand what he was talking about.

One thing I wondered about was he seemed to
imply that he/they invented 5.1 sound for "Apocolypse
Now".

Yeah some if it feels far out there, but I guess that's the nature of working with sound. Another interesting read is Randy Thom

If you just want a quick overview Robert Bresson's Notes are much shorter.

One thing Murch talked about was "dynamic range".
I think he said television dynamic range is only
+3dB and -3dB from an average dB point.
Anyone know if this is true?

He also said that, for many years
(like the 1930s to the 1960s),
cinema movies were limited
to +6dB and -6dB from an average point.

Well, having worked at TV stations I'd say it's mostly true.

For a live instudio production, I'd send an uncompressed feed to master control, it would go through two stages of compression before it hit the transmitter, which also messed with dynamic range. Some strange stuff came out the other end.

What was amazing was that the off-air recording compared to a recording of the cable feed compared to what we were sending - all very different, and a mix for one output would sound terrible coming back from another output. (I could monitor the various outputs from my console).

Many times I was told "don't worry about this - just make sure what you send is good." Hard to judge my work a success in that environment, I don't work in TV any more in part because that really bothered me.

Remember, most people listen to tv on a 3.5" speaker or two in a lousy cabinet.

This makes me think a little about some of the podcasts I've heard from NPR. They are obviously pre-compression signals, not the signal that was actually broadcast. I can barely listen to them in the car or at home without riding the volume up and down because of the massive changes in dynamic range (some guests talking right into mics, some guests talking off to the side, etc.).

Well, I make DVDs for playing on home TVs and I've
basically been keeping the levels between +6 and -6
from an average point. Usually I need one or two
stages of compression to get it there. Sometimes
I am not happy with the effect a software limiter
has on the sound because you can hear it clamping
down. I've tried making a longer attack time, which
has helped some but ...
I'm using Vegas and I wonder if there might be
other software that would give better (more natural)
results.

One thing I'll keep in mind is Murch said not to use more
than 5 tracks at once. He said dialog track is most
important. The other tracks are music, room tone,
special ones like footsteps/door knocks. The last
track I wasn't real clear about but I think he
was saying it was an overall general track that
included all of these, so I'm thinking maybe like
he's saying here one mic that takes it all in.

If you bought Sound Forge, it probably came with Wave Hammer which you should check out - it's more of a volume maximizer (which is good).

A truly outstanding plugin is Ozone from Izotope, it's a complete mastering plugin with 6 or so modules, including a great volume max as well as a multiband compressor, and others. I use it all the time for audio and audio for video production.

I think some of Ozone (including volume max) is bundled in with the latest version of Ultimate S from vasst.com. I believe there's a trial version available.

To reduce "pumping" with a conventional software or hardware compressor, I usually use an extremely short attack, a moderate ratio of perhaps 2.5:1, and a release of 1-1.5 seconds. Then, I reduce threshold to where peaks are getting from -3 to -6 of reduction. This isn't classic use of a compressor, it's more like a soft limiter.

Of course it really helps if there isn't much background noise, clean recordings show fewer pumping artifacts when compressing.

So, I might do that at a track level, depending on the content, then strap a volume maximizer across the master output.