Square One: Audio Poltergeists

Image placeholder title

The venerated sage Murphy declared that anything that can go wrong will, and tradition holds that this will occur at the most inopportune moment. Smithers's Corollary to Murphy's Law asserts that the only defense we have is to attempt to understand every type of problem that can be anticipated and have a plan to prevent it and a contingency plan for when it happens anyway.

With that in mind, here's a laundry list of pesky poltergeists ready to ruin your audio adventures. I won't discuss distortion problems (except for digital clipping) and A/D conversion issues because those were discussed in recent “Square One” columns (see “Sonic Mayhem” in the August 2007 issue and “Preaching to the Converted” in the June 2007 issue; both are available at emusician.com).

One Foot in the Ground

For electronic musicians and audio engineers, squeaking chalk is nowhere near as annoying as 60 Hz hum. Two common causes of this pesky noise are ground loops and induced hum.

Ground loops occur when electricity has more than one path to ground, creating noise in the signal path of the equipment. Because AC power is delivered as a 60 Hz sine wave in the United States (50 Hz in Europe), the noise is heard as a 60 Hz tone. The best fix is to avoid or remedy the ground loop, which is easier said than done because ground loops can form in a number of ways. Use nylon hardware to mount metal equipment cases to metal racks, and avoid metal-to-metal contact between pieces of gear. You can “lift” the ground by disconnecting the shield at one end of an audio cable or by using the ground lift on a DI box, but this is not always the best approach, and you should never lift the ground on a power cord.

Induced hum is most often caused when audio and power cables run alongside each other. The magnetic field surrounding a power cable induces current into the audio cable. Although balanced audio cables are better at resisting this type of interference, the best solution is to keep audio and power cables apart; if they must touch, cross them at a right angle.

Hum can also be introduced by air-conditioning units, refrigerators, and lighting fixtures (especially lights with dimmers) that are on the same circuit or ground plane as your studio. A good power conditioner with filtering can manage many such problems, but this is much like fixing bad tracks in the mix — you're better off avoiding the problem to begin with by putting your studio on a separate AC circuit with an independent, true-earth ground. (For an in-depth explanation of grounding issues for audio gear, including true-earth grounding, see “On Solid Ground,” originally published in two parts in the September and October 1992 issues of EM and available at emusician.com.)

Sometimes audible hum is caused by the inherent noise of a component, made louder by poor gain staging. When several devices are connected in series, such as a microphone into a preamp into a compressor into your DAW, it's essential to optimize the signal-to-noise ratio at each step. If the preamp is turned way down, the makeup gain on the compressor will have to be cranked up to compensate, which amplifies the inherent noise of the preamp along with the signal. It's best to set the preamp loud enough to maximize its signal-to-noise ratio so the compressor doesn't boost the noise floor.

Binary Bugaboos

Image placeholder title

FIG. 1: Digital clipping is as obvious to the eye as it is to the ear. The characteristic flat-topped waveform usually means that the signal coming from one device overloaded the input stage of the next device in the chain. Repairing the clipped peak requires specialized software.

The most obvious problem that crops up in digital audio is rapidly becoming a deliberate effect, even a trademark sonic signature for certain engineers. I'm referring, of course, to clipping, the flat-topped waveform that occurs when the signal level exceeds the system's headroom, most commonly at the input stage (see Fig. 1).

If you overdrive an A/D converter, you will hear distortion ranging from clicks and pops to static to edgy harmonic distortion. Unlike analog circuits, which usually overdrive gradually and clip forgivingly, digital devices simply run out of numbers to describe an excessive input level, which sounds nasty. Clipping also occurs when you overdrive a DAW's mix bus or D/A converters.

The solution is to turn it down. Lower the preamp's gain; turn down your source tracks or master fader. Know your device's meters and respect what they tell you. If you still find yourself with a clipped waveform, a couple of fixes are available. I once salvaged a live brass quintet recording by carefully lowpass filtering a clipped phrase. Because the squared-off top of a clipped waveform resembles that of a square wave, it creates similar high-frequency components. Filtering the signal rounds off those corners much like the smoothing (antialiasing) filter of a D/A converter does to a stair-stepped PCM output signal. More-sophisticated repair is available from audio-restoration software, such as the declipper found in iZotope's RX suite. But it's better to avoid the problem to begin with.

Time to Worry

Image placeholder title

FIG. 2: SoundHack is a venerable freeware program that offers many useful, if sometimes obscure, functions for operating on audio files.

Several types of timing problems can be hazardous to digital audio. The most obvious timing problem is a file with the wrong sampling rate. There are various ways this can happen, most of which boil down to somebody or something telling the DAW one thing and the master clock another. In this way, a file gets recorded at an actual speed of, say, 48 kHz when its file header identifies it as a 44.1 kHz file. Import it into another session — or play the original session back with the master clock set correctly — and it will sound almost a whole step lower than it should. The easiest fix is to edit the file header to reflect the proper sampling rate, using software such as Tom Erbe's SoundHack (soundhack.com; see Fig. 2). In some DAWs, you can accomplish the same thing by importing the file into a 48 kHz session without sampling-rate conversion, then importing it back into the 44.1 kHz session and letting the DAW convert the sampling rate.

Jitter is the inherent irregularity of a clock signal, and it causes converters to sample earlier or later than they should. Jitter in A/D conversion captures that distortion to the file, whereas jitter in D/A conversion has no permanent effect on the file or sessions being played back. Use high-quality converters, and use a stable word-clock source when synchronizing multiple digital devices. If you have no master clock and are simply daisy-chaining the word-clock signal, use your A/D converter as the master clock when recording and your D/A converter as the master when mixing or printing a mix.

Jitter between digital devices is irrelevant as long as all devices follow the same clock. If there is any confusion about the clock chain, however, samples will be dropped or doubled as clocks drift apart. Use the correct cables, terminate the end of the chain, and learn your gear's temperamental side. When rational analysis fails, don't be afraid to move past what should work and experiment with different chain orders.

Ex Files

Applying lossy data compression, such as MP3, inescapably reduces the audio quality of a sound file, and converting the file to an uncompressed format such as WAV does not undo the damage. A less obvious artifact of data reduction is a tendency for a waveform to drift out of sync compared with the original. Trying to remarry a soundtrack to a video when each has been compressed can be frustrating. Careful editing and judicious use of time compression and expansion can save the day. It's better to use the uncompressed originals if they are available.

Sampling-rate conversion is sometimes a necessary evil, but it should be handled with care. Always use your DAW's highest-quality conversion algorithm. When possible, don't convert between multiples of 48 kHz and multiples of 44.1 kHz, as the required math is complex. (For an explanation of the math involved in converting between 44.1 and 48 kHz, see en.wikipedia.org/wiki/Sample_rate_conversion.) I may say “24/96” for simplicity, but I actually record CD projects at 88.2 kHz. If you can, leave the sampling-rate conversion to the mastering engineer.

There is, however, no sonic difference between the various PCM formats. You can convert from WAV to AIFF to SDII and back to WAV and never alter the sound.

Keeping Murphy's Law at bay means using the left side of your brain to manage your right-brain activities. Plan ahead for your creative sessions so you can face the music with as few distractions as the law allows.

Brian Smithers is department chair of workstations at Full Sail University and the author of Mixing in Pro Tools: Skill Pack (Cengage Learning, 2006).