Live and Vocal

by Steve Dove
Wheatstone Minister of Algorithms

Steve DoveThere's a big difference between what it takes to get live voice straight to air, and what the sound engineer needs to do for audio that will be post-produced. In the latter case, it’s always a good idea to just concentrate on getting it all down cleanly with good consistent levels and minimal processing. The boys in post-production will definitely not thank you if they have to try and unwind heavy EQ you wound in, or deal with irreversible deep compression.

Ah, but live... let's talk about getting the glinty-teethed talent behind the desk sounding right.

First, go into the studio and hear what they actually sound like, both their normal conversational voices and their "on" persona. This is your target, not some arbitrary notion of what they ought to sound like.

Microphone techniques in TV are, charitably, non-optimal and driven by the visuals. Now, tie-clip mics actually sound a lot better than we have a right to expect but they are (usually) omnidirectional, poorly located on the chest, tend to hear a lot we rather they didn't, and what they do hear is colored. Over the years the mic manufacturers have attempted to mitigate their shortcomings, but there's still work you can do with tools to hand. A modern digital TV audio console such as a Wheatstone console has all the necessary audio tools on board, equal to or superior to those found in the best recording plug-ins and the like. They're there for good reason, so let’s get started.

Get Rid of the Garbage

Set the console input gain conservatively, so that there's headroom if the presenter gets excited. It may be wise here to set the channel's compressor to start acting at just about “normal” voice-level to catch anything louder (say, 4:1 ratio, 50mS attack, 300mS release) yet stay transparent. It's for protection, not effect.

As a matter of course, dial in the high-pass filter to, say, 50Hz on a male, 80Hz on a female presenter -- this will remove a lot of untoward noises without affecting their voices at all. Microphones reveal just how noisy studios actually are, though they seem quiet to us. Adding an expander (think: subtle noise-gate) to each microphone not only reduces background noise when the presenter's not speaking, but reduces coloration between multiple live microphones. The expander should be set to open only on that presenter's voice - not his neighbor’s. (Open, fast <1mS; hang-time, say, 100mS; close, 100mS. Other manufacturers may use different nomenclature, like “attack,” “hold,” and “release”). Most importantly, it's only necessary to use very shallow gain-reduction -- a handful of dB -- to make a world of difference. Full gating is clunky and obvious.

So far we've dealt with protection (compressor) and garbage disposal (high-pass, expansion). Now, let’s get into qualitative sound.

Reduce Clip-On Mic Chest “Resonance”

Typically, a clip mic hears chest resonance -- which is actually a bit of welcome low-frequency energy -- but it is a bit “'weird”' sounding. Using a band (or two) of low-mid parametric EQ can help bring this oddness closer to sounding like something human. The main issue is usually a kind-of “one-note-bass”' resonance; use a parametric section to isolate this frequency and reduce it.

(Only do the following during sound-check -- it'd be a gigantic bust live on-air!) A good, fast technique for tuning in on a problem such as this is to set the parametric's “Q” to, say, 3 (bandwidth 0.3 octave), turn up the section level to full(!), and tune the frequency until the issue leaps out at you; turn the level down to unity, change the “Q” to 1 (bandwidth 1 octave), and carefully start attenuating the section, just a dB or so at a time, until the resonance is less obvious. The voice will sound a bit thin at this point -- you are, after all, removing low-frequency energy, if pinpointed -- so carefully bring the bass back up using a low-frequency shelving EQ until the voice sounds more in balance. This should help the bottom end of the voice sound more natural; at least it'll be a good start so you can gently nudge around parameters to tune-in more finely, of course resisting the urge of endless beautification that can obsess us all.


Solve Sibilance

Voice-ProcessingToolsOfTheTrade 2560In order to enhance clarity and articulation, clip mics are very “bright,” meaning they make a lot of high-frequency energy to compensate for the fact that they're a long way from the mouth and way off axis. The good news is that this saves you from having to wind in a lot of HF EQ. However, you have a bit of work to do to make them sound natural, and at worst, you might be dealing with an overly-sibilant voice that can bring down aircraft.

Depending on the mic, this HF energy can be “peaky,” with a classic articulation boost in the 3 - 8kHz region. Using the “search and destroy” technique described above, you can dial in a relatively broad EQ to change the peaky enhancement to a gentler, smoother sound. Alternatively, start with a parametric section set up for 5kHz, Q 0.5 (bandwidth 2 octave), and gingerly attenuate. If this doesn't allow you to tame the “shriekies” without the voice sounding dull, then the next weapon is the de-esser. This is a dynamic equalizer that controls the maximum signal level within a chosen piece of spectrum. Sibilance is, unfortunately, highest energy just where the microphone's response peak is greatest; this is where the de-esser's frequency is set - a good start is 5kHz. The de-esser's sensitivity is adjusted to best control the sibilance by ear. The magic is that unlike ordinary EQ, the de-esser leaves the spectrum completely intact when it's not actually “biting,” so it doesn't sound dull.

Round Out the Voice

Now that we've dealt best as we can with the shortcomings of clip-mics and their placement, it’s time to round out the voice. Start with “gentle” EQ to try to get the vocal as close to natural sounding as possible. There shouldn't be any wild EQing necessary.

Clip-mics aren't the only things you'll come up against. Often there'll be a pencil microphone as a backup, typically placed below desktop level out of eyeshot. There may be a boom-mic -- a shotgun-on-a-stick -- above the presenter's head. Needless to say, these will all sound completely different “raw.” For continuity's sake, your job is to make them all sound convincingly similar to the primary microphone. The same basic housekeeping processing applies (high-pass filter, protection limiting/compression, noise-gate/expansion). To achieve anything close to the same level of enhanced articulation, a fair amount of high-frequency EQ may be necessary. Perversely, the improved low and low-mid characteristics of these mics may need tweaking to match the less-accurate and more contrived sound of the clip-mic. The pencil mic, sub-optimally surrounded by reflecting surfaces, will need quite a bit of “search and destroy” EQing to straighten out. Remember, consistency is what you're trying to achieve.

Many presenters just LOVE the way they sound with deep compression, and may bug you for it. Something to bear in mind before you bring down his compressor threshold from “safety” to “he-man”: When the audio leaves you, the chances are good it may be going through an airchain processor of some sort, even if it is a relatively gentle loudness controller. This will inflict a measure of compression almost by default. Unless, managed carefully, cascading compression rarely works out well.

If he insists, you can always try compressing his IFB feed. (Not that I’ve ever done it.)

Site Navigations