Skip to main content

Gain Staging The Most Important Thing In Audio?

As the world goes increasingly digital, one problem potentially recedes from importance, though its effect is still being heard on news programs all the time: high distortion from clipping. When a program contribution has been clipped, which usually involves a gross timbre change of vowels in particular, a wide frequency range monitor loudspeaker setup shows the sound is awful. Perhaps a small 3-inch monitor speaker somewhere doesn't show this as much due to its limited high-frequency response, and the editor thinks things are all right, but played over a wide range monitor in the studio, or increasingly, in the home, the problem is obvious. In Hollywood, this syndrome leads to one of the most common sayings by editors: "It sounded all right on the Avid!" This is meant to imply that the sound people are far too picky, and how could it be so different where they are listening? Well, when one of these events happened, a sound technician went into the edit bay and found all the tweeters in the monitor speakers were burned out! Adding to the editor's misperception of the audio quality is the high noise of hard drives and computer fans in practically all such environments. It never fails to amuse me to go into a telecine suite and find that the noisiest thing in the room is the digital audio monitor! Just what do the designers think the user is doing?

If much of the path is digital, there is not much likelihood of a unintentional level shift in the digital domain clipping the signal. However, we still have analog parts of the path, and here gain staging is all important. I'm sure that the reason for most of the distortion we hear is this: that the level somewhere in the chain not being monitored is clipping, while at the metered point in the chain, the level is reasonable. After all, for instance, we've got professionals running these cameras (we hope), who monitor "level," no?

Let's take a setup with a Betacam SP camcorder and an electrostatic (the proper name for condenser or capacitor) microphone. These types of microphones are highly prized for wide frequency response and greater sensitivity than electrodynamic types, so they have greater "reach" because, due to their greater sensitivity, you can capture quieter things with them. This also means they are "hotter" on their output, and more likely to clip microphone inputs. One example is of a Sennheiser MKH416 short shotgun plugged into a mono Nagra recorder (OK so you don't use these in news, but I've got the numbers on them). The preamplifier is designed for dynamic mics. To pad the clipping voltage level of the microphone (1.25 Vrms) down to the clipping voltage level of the preamplifier (20 mVrms) requires a 35-dB pad! That's not a typo and it isn't missing a decimal point. It is what you need to record the full high end of the dynamic range of the microphone on that recorder.

I give a talk to my students called "Cries and Whispers." That's a tongue-in-cheek reference to Ingmar Bergman's film, but it's also a reference to the fact that sound people are expected to record both ends of the dynamic range. When sound was introduced to the movies, a "sound director" made the actors perform at a consistent level. In recent years it has been the case that if the director standing next to the camera can't even hear the actor, it is still expected that the sound person will be able to capture this subtle performance. Stand up news people are a whole lot easier, but what has happened is that the most sensitive mics have won out because they're hotter (and that's always better in a comparison). It also means they are more likely to clip the microphone input, that may have been designed for the classic EV 635 dynamic mic with a sensitivity of 1.9 mV/Pa (1.9 mV at 94 dB SPL). Now we connect a Sennheiser, with a sensitivity of 25 mV/Pa, and the microphone itself is 22 dB hotter!

That's where the problem comes in. Because ENG camera people are only trained on the metered level, which is after the microphone level control, and since they are monitoring over the tiny loudspeaker built into the camera if at all, they don't know that the audio is badly clipped at the input to the microphone preamp. They also don't know that if they are using an electrostatic microphone, they will almost always need the pad. There should be an alert clipping flasher in the viewfinder on the analog audio electronics, but there isn't. That's probably why we hear so much bad field audio today.