So, the shoot is over and the tape is in the case. For better or worse, you have to work with the audio captured on location for your final edit. If you were lucky enough to have other folks shoot the footage and capture the audio, you can curse their names with disdain as you pour over the footage in your edit bay. If you are the culprit with the camera and mike, close the door, watch and listen to what you’ve got, and take your lumps. Go ahead and wince, it’s expected.
Level with Me
Most likely, you edit your video on a computer. Most software plug-ins and mixing tools emulate the tried and true techniques and devices of the old analog path. For argument’s sake, when you digitize or capture your footage with audio into your computer, the transfer is the most important step. With audio, you want to transfer the strongest signal possible onto your disk, short of clipping the meters (usually indicated by the red light that stays on at the top of your level meter display).
If you capture via a digital IEEE 1394 (FireWire) connection, this is largely a moot point. The transfer will simply transfer the data stream from your tape to the hard drive, so whatever is on your tape will end up on your disk.
If you digitize your video through an S-video cable and your audio through unbalanced RCA cables, you have some choices to make about how you will treat that signal as it enters the digital domain. A direct transfer is easiest, straight across from video camera to digitizing capture card. Software programs usually have default settings for input levels, but you can raise or lower that value if your tape signal is too high or too low. If your clip peaks every once in a while during transfer, you shouldn’t worry. What you should avoid is hitting the clip continuously or having the level set so low that it has no chance of lighting the clip. In digital audio, the level, or amplitude, is a function of "bits." If you use 48kHz, 16-bit audio, then the higher your meters read, the more bits you are using, and the more well defined the digital representation of your audio signal will be.
Also, if you digitize your audio, you may have the ability to run it through an audio mixer and adjust the signal in the analog pass, if you desire. But, I do suggest that you make adjustments conservatively when going through your mixer into your computer. In most cases, limit the mixer’s role in the transfer to riding the levels in areas that you know are problematic, such as a loud noise of a truck, or using the EQ section to roll off offending low frequencies, like wind.
Ups and Downs
In the editing process, as you assemble your audio and video tracks onto a sequential timeline, you will see the graphic representation of the waveforms in your recorded tracks. The noise floor, or the ambient noise of your audio recording environment, will show up as a constant wave.
You can go a long way adjusting the gaps between words and phrases with the rubber-band volume controls in your editing software, mitigating the noise when possible with simple level adjustments. But lowering a constant noise like machinery too much or cutting it out completely can be more distracting than the sound itself. So, when lowering the noise in the gaps, you must do it with care. In fact, if you need to cross cut footage that includes background noise with one that does not, a sample of the noise can be placed under the noiseless clip to smooth the abrupt change between shots.
Pass it, Notch it
The good thing about constant noise (if there is a good thing) is that it has a distinct wave signature. Wind creates a low rumbling sound as it passes across a mike, machinery often has a constant drone and the 60-cycle hum from electrical interference is, well, 60 cycles or Hertz. Sound that generally holds wave shapes like these can be processed and dropped significantly using tools that come with almost every video editing application. High pass and low pass filters can cut out a large portion of the overall sound.
When a high pass filter, for example, is set at 100Hz, every sound below that frequency will drop off. This can make wind all but disappear and does little to harm the voice frequencies. The converse is true of the low pass; it shuts off the shrill and hisses of the audio spectrum when set to frequencies above 8 to 10kHz. A notch filter, on the other hand, hones in on a narrow offending frequency anywhere in the audio spectrum and eliminates it. The notch will drop out the chosen frequency range and leave those close-by undisturbed. These can be valuable tools that can clean up a track without much adjustment.
Can’t Beat it? Mask it
One good thing about the human ear is that it can be fooled. When you add music and sound effects to the mix, you can hide a world of faults. After you have fought the good fight and adjusted levels, smoothed transitions with ambient noise and cut what you dare with EQ filters, add some music to mask and blend. Music under a track not only can change the emotion of a scene, it can hide a fair amount of offending noise.
Some musical textures will work better than others in certain situations, so listen to determine which fits your video’s mood and helps to blend with your sound track. For instance, if you produce a show about race cars, you may have on-the-track interviews where the participants have to shout to be heard above the din. Using hard, pounding music will fit the mood and help the listener to make transitions between takes without violent swings in audio levels. On the other hand, if you have a quiet interview with a soft-spoken senior, the hiss of the electronics in your camera may be enough to drive the listener out of the room. Soft strings and gentle melodies could be just enough to avert the attention from the roar of that hissing "silence."
And Then There’s Magic
In the early 1990s, the professional audio world came up with software tools that were, to say the least, wondrous. Plug-ins, like DINR (digital intelligent noise reduction), from companies like Digidesign, took noise eradication to new dimensions. What DINR did was sample the waveform of the noise in question, say, the sound of air-conditioning, and create a profile of that wave signature. After the noise was profiled, the software eliminated all the data in the digital file with that particular signature. With DINR, you could remaster and clean up pops, scratches and hiss from old music recordings and movies. It was a very sophisticated tool that was also very expensive, but now, other companies make more affordable (sometimes less than $200) variations of this same technology. A quick search of the Internet will turn up many and it shouldn’t be long before many audio and video editing interfaces come with a host of noise killers. But don’t forget the basics of good sound production, because those techniques will never go out of style.