An introduction to the thorny subject of phase (without using maths).
In 1710, philosopher George Berkeley asked, “If a tree falls in a forest and no-one is around to hear it, does it make a sound?” This was a thought experiment about perception, observation and reality; if nobody perceives the sound, is it really there? We can comfort ourselves that if a tree does indeed fall in the forest, it only has to worry about being in phase with itself. To awkwardly amend the beauty of the question, I’d like to ask, “If two trees fall in a forest and no-one is around to hear them, are they in phase?” It may seem glib, but it highlights the importance that the phase of a signal should be considered relative to the phase of another, unlike timing where two things can happen at the same time but not be at the same point of the wave cycle (phase). Phase differences are also not fully solved by our perennial fall back of “flipping the phase” at the top of the channel strip either; this simply makes a signal more or less out of phase with another as it is a simple 180˚ polarity invert and not a tool to line them up fully. Phase is also relative to frequency.
Why should this be important to you?
It has always been important to us. SSL is known for producing audio equipment with exceptionally good phase response. By minimizing the amount of timing differences that the components add to signal passing through them, we bring all frequencies of every signal closer together in time (or more accurately, I should say, “less out of time with each other”) and minimize any changes in phase due to processing. This leads to clearer stereo imaging, better separation and a fat deep low end. Our SuperAnalogue circuitry is designed to minimize micro-shifts in phase that you find in electronic designs that have electrolytic capacitors in the audio path. These lead to time smearing across different frequencies, and our Mix Buses are renowned for their phase cohesivity.
I’d always thought that the most important things to remember about phase were measuring distances from source to microphone, and lining up the timing in my workstation after recording. As audio engineers, we probably already have an idea of how phase difference between multi-miked sources can play merry hell with the final result. It is not a difficult concept to understand that two microphones picking up the same sound source but positioned different distances away from it may not line up right, and mixing them together can lead to comb filtering of the signal. Quite why I never took the leap to think that it is also important to consider any phase changes introduced because of the equipment you run the signal through, I’m not sure….
SSL’s X-Phase plug-in. Pretty much anything you run a signal through is likely to change the phase of different frequencies within it to some degree. For example, a high-pass filter will introduce a phase shift of 90 degrees for each order (a 4th order high-pass filter will therefore introduce a 360-degree phase shift) on the frequencies below the filter frequency. If you strap a high-pass filter on a bass track to make space for the kick drum in a mix, then any remaining frequencies below it in the bass part will be 360 degrees out of phase with those frequencies in the kick drum. Hello again, my old friend notch filtering! When I started digging into it a little deeper, it astounded me to discover that simply running signals though certain equipment that is widely considered to be ‘high end’ (no names mentioned) can introduce a -70˚ phase shift in the bass frequencies and a +70˚shift in the high frequencies of the signal at the same time. How would I handle how this might affect a parallel of the same track once you try and combine them?
I think we all get a little daunted by the maths involved with phase and often settle for “lining up the audio tracks in my DAW” thinking that this will get the job done. Recently I’ve taken a different approach and have started using an All-Pass filter (specifically SSL’s XPhase plug-in) on many of my tracks in a mix. Unlike an EQ, which changes the amplitude of selected frequencies in a signal, X-Phase keeps all frequencies of a signal at equal amplitude but changes the phase across all frequencies. A phase flip is performed around a selected frequency with a variable slope around that frequency. Keeping things simple and practical (not getting an Audio Precision out and measuring the phase of things), I just started to sweep the All-Pass filter across a particularly tricky bass that had been miked and DI’d and was astounded to hear how it focused the sound and made it sit better in the mix. I started to do the same on the drums, and before long, there was an “unveiling” of the individual tracks inside the mix. Granted, it was not the kind of difference that some who cannot differentiate between a .wav and an .mp3 would hear straight away, but it was certainly audible to me. Perhaps it is time that we started to take phase a little more seriously….
Jim Motley is Head Of Business (Workstation Partner Products) at Solid State Logic.
Solid State Logic