Some of the more alert readers among you may recall that in my August column I wrote about the low-frequency effects channel in multichannel video and music production, and made some observations about how it fits into the end-user environment. I took some audio signal level and spectrum measurements of two commercial film releases on DVD--"Das Boot" and "Shaft"--observing the various channels to show something about the relative levels and spectrum of the main, surround and LFE channels.
Dave Weinberg, publisher and editor of the Boston Audio Society magazine "Speaker," and a quite dedicated and knowledgeable audiophile, wrote, "It is my understanding that the LFE levels as recorded on transmission media, be it DVD or film in Dolby Digital, DTS, or SDDS, is designed to be recorded 10 dB low and boosted during playback by 10 dB... Thus the relative levels you mention and show in your graphs for the LFE are 10 dB lower than playback sonic level in the room."
Oops! I had measured levels directly from the DVD player's analog outputs, referenced to 0 dBV. In production, I send all channels to the monitor console at "unity" gain (I don't need a subwoofer so I just send the LFE channel to left and right). Being an old-school kind of guy, it never occurred to me that those DVD channels would be different, although I've heard mention of the idea that the LFE is down 10 dB.
So, I wrote back to Dave, saying, "I'm aware, of course, of the notion that we should raise the level of the LFE channel by 10 dB..." I went on to argue that my measurements still showed what I said they showed, which was that the subwoofer signal content wasn't a very important part of the home theater audio experience, relative to the spectrum and amplitude of the main channels.
To which Dave replied, "The 10 dB LFE boost is because the recorded levels are dropped by 10 dB for the soundtrack recording and transmission to the theater. If you want the same relative levels with the other channels as originally created, then the LFE needs to be boosted on playback, which consumer processors handle automatically. In my Lexicon, for example, with the band-limited pink noise level calibration signal, the directions are to set the measured level to 75 dB SPL from each channel, including the subwoofer channel. The processor ensures the LFE level is properly adjusted to match the original mix as transferred in DD or DTS onto the DVD."
This got me to thinking. Why wouldn't there be a level correction inherent to the DVD and the DVD player? Why would we configure a consumer system to have a "wrong" level on playback, and to depend on a consumer processor to automatically make up that level error?
WHAT DOLBY SAYS
So I called Steve Venezia, manager of DVD/DTV broadcast support at Dolby.
Here's the gospel according to Steve, and supported by the Dolby 564 decoder manual. When a production monitoring environment is calibrated, the subwoofer itself is calibrated so that a band of noise between 20 and 120 Hz sent to it is measured to be 10 dB greater than the reference broadband level measured at each of the main speakers. These are each typically set at 85 dBC SPL. Note that it is the subwoofer that is being calibrated, not the LFE channel (which is not adjusted), and it as measured using a real-time spectrum analyzer driven by a microphone at the mix position. With that calibration in place, we mix to taste, including the LFE channel, and no further correction to level is done in the production chain.
It stands to reason that when we want to listen to said production, we'd like to have a similar setup. No problemo. All we need is to own all the appropriate test gear and have the requisite skill set and knowledge base! But what's really interesting is that the power of that narrow band of frequencies for the subwoofer is highly dependent on room acoustics and the efficiency of the subwoofer itself. A simple 10 dB change in level sent to the subwoofer isn't going to do what we want. It is probably too much boost. In fact, the 564 manual says, "If an RTA is not available, setting the subwoofer channel 4-6 dB high, as measured by an SPL meter, provides an approximate level. For example, set the subwoofer channel to 89 dBC SPL when the center channel measures 85 dBC."
FOR THE END-USER
So, once again, for reasons that are sweetly reasonable in production, we've gotten to a level of complexity that is going to be beyond the end-user.
Joe and Jane Sixpack are not going to own an RTA, a variable bandwidth noise generator, an SPL meter, a test mic or an audio engineer! Therefore, they are not going to be able to configure their home theater correctly. They'll have to set the subwoofer level by ear, with no meaningful references.
CEDIA-certified installers have a slightly better chance of getting it right, because they go in with a set of instructions and some instrumentation. Hopefully, the instructions they get from the manufacturers are close to correct. Unfortunately, what Lexicon apparently told Dave Weinberg doesn't sound quite right.
LFE FOR FUN & PROFIT
Steve Venezia also pointed out to me that it is reasonable to think of LFE as standing for "low frequency enhancement" rather than "low frequency effects." It is there, in film and video, to fill in the bottom end for sonic FX; It has little place in music. From my standpoint, as I said in the August article, the use of the subwoofer as a cost-cutting way to use five small satellite speakers is a usage that we can't reasonably produce for (although we would be wise to check our work on same). As an enhancer to supplement full-range speakers and give us just a little extra oomph each time the world ends, the subwoofer works fine, particularly if we approach it conservatively. Thanks for listening.
Dave Moulton would like to thank Dave Weinberg, Steve Venezia and Jeff Riedmiller for their attempts to keep him out of even worse trouble.
