Dolby Digital center downmix - Does it work?

[Martin Pauly]

I have a 5.1 mix of a comedy show that my customer and I are very happy with when played on a 5.1 audio system. The center channel has the stage mic, the other four channels carry sound recorded by different microphones placed in the club.

We've also auditioned it on cheaper TVs, where the DVD player does a downmix to stereo, and it doesn't sound that good. Not that we would expect the same quality as on a high-end 5.1 system, but the problem is that the audience sounds too soft in the downmix, which I assume is the same as saying the stage mic is too loud.

During the AC3 encoding, both center downmix and surround downmix where originally left at the default value (-3dB). So my idea was to lower the center channel in the downmix to -4.5dB or -6dB - since there's no option to raise the other channels - hoping that overall that would give the L/R/Ls/Rs channel more emphasis (relatively speaking). The thing is, no matter what values I choose, it doesn't seem to make a difference when I play it on my cheap DVD player and TV.

Is this something that not all DVD players will support? Could this be a problem with the encoder (Apple Compressor)? Would the idea even work? Or would I be better of creating a second audio track using a stereo mix that I can create and thus control better? (I'd like to avoid having the user select between 5.1 and stereo if possible.)

Thanks in advance for your comments!

- Martin

[Steve House]

You said your L/R/Ls/Rs channels were recorded from "recorded by different microphones placed in the club." This sounds like your L and R channels were not recorded with a true coincident stereo mic arrangement but instead were done with randomly positioned mics separated by some distance. I suspect what's happening is the time differences between when a sound arrives at one mic versus when it arrives at another is creating phasing issues and partial cancellations between both the L and R front channels and the fronts and surrounds when downmixed. You may have to bite the bullet and create that separate mix for this program.

You might be interested in this white paper on the Schoeps site regarding mic placement for surround recording. http://schoeps.de/PDFs/oct-e.pdf

There's also a good summary of the pros and cons of different approaches to surround in their current catalog.

[Seth Bloombaum]

Steve could be right.

Hard to tell without listening to what you're hearing.

I'd also suggest looking at the whitepapers that are available on Dolby's site. This situation is ringing a bell with me but I don't have time this morning to look at it - check out the "dialog normalization" parameter.

A cheat that might work (depending), is to create a left-right stereo mix in AC3 instead of 5.1. Working a mix in stereo should help you to figure out what's going on. Also helpful to monitor in mono before going to DVD.

Are you able to monitor both in 5.1 and in stereo when you're mixing this? That's very helpful.

[Martin Pauly]

Thanks, Steve and Seth. I see your points with regard to the recording. I read the Schoeps article; my mic placement is very different from their recommendations, but I think it should still work in my case. Here is why: the center channel is tied to the stage mic, which in this case was a dynamic mic (SM58) that picks up just the comedian, and almost nothing else. Four other cardioid mics were placed high above the audience, pointing down. Because of the distance between mics (I'd say > 20 feet) and the inverse square law, I don't expect any significant overlap between the four audience channels, and monitoring each channel separately confirms this: they pick up a similar sound from the PA, but very different portions of the audience response.

The perception that I am aiming for with my surround mix places the listener in the front section of the audience, hearing the clear, understandable voice of the comic in front of him, other audience members all around him (incl. people sitting behind him), and some of the PA in the room. To me, it does sound pretty darn good when played on a 5.1 system. I do agree, though, that my setup would be a poor choice for other goals, such as capturing the room characteristics in a concert hall while an orchestra is playing - the Schoeps recommendations make a lot more sense there.

But assuming that you guys are right, and that phasing issues and partial cancellations are causing the stereo mix to suffer: shouldn't the center/surround downmix parameters still have a noticable effect, even if they may not be able to completely fix the problem?

I did search some more, and found a discussion on Apple's support forums where someone posted a question that sounds a lot like my case: the downmix parameters seem to be ignored:
http://discussions.apple.com/message...sageID=6918034

Are you able to monitor both in 5.1 and in stereo when you're mixing this?

Not really. I am mixing with Apple's Soundtrack Pro 2. The manual has a short section about "Accommodating Stereo Playback" (Chapter 9, page 312), but I haven't found any tools inside this application to simulate the AC3 decoder's downmix, with the center and surround downmix parameters of my choice. But the problem of these parameters apparently being ignored would still remain.

I think I'll just create a stereo downmix by hand for this project - but I am still very curious what's up with the AC3 downmix parameters.

Thanks again for your input!

- Martin