Zero Sum Gain

Most existing methods of stereo imaging and positioning suffer from unfortunate weaknessness. Gain panning is of course the most simple to implement and is ubiquitous throughout the last 100 year history of recording arts and sciences. Unlike some of the more modern methods, it is generally free from any nasty unwanted side effects. There remains one rather large issue however: it generally does not occur in real-world acoustic environments -- at least not to the degrees and intensity offered and commonly used by mix desks and modern DAWs. A hard-panned signal is not particularly natural especially on headphones for example. Drastic level differences between ears, or stereo microphones simply almost never occur in the real world. In the real world the only equivalent would be something like a young lover whispering directly into only one of your ears, and while this may seem nostalgic indeed, it simply does not really translate to the experience of musical instrument performance in acoustic environments. There is almost no desirable circumstance in which someone could hear a violin, trumpet, piano, drums, or vocal in only one ear. Of course play-back of gain-panned signals in-air over loudspeakers results in sound reaching both ears even if sound is emitted from only one of the loudspeakers. Some mixing in-air certainly occurs, but not necessarily in physically accurate ratios relative to the original sound field. More problematically this method completely ignores other important interaural differences that occur in nature. Gain panning is used widely because it is mostly safe to use, not necessarily because it is the best choice aesthetically or provides the most accurate directional cues. Differentiating a large number of instruments in a mix based solely on gain panning is extremely difficult and often results in poor separation between instruments.

Haas (The Precedence Effect)

In the real physical world a sound coming from a source that is closer to one ear or microphone will arrive at this ear or microphone before the other more distant one. In technical terms this is called an Interaural Time Difference and in human perception this cue has a significantly stronger perceptual impact than Interaural Level Differences (i.e. gain panning). In psychoacoustics this phenomenon is called the Precedence Effect or the Haas Effect as named after one of the first scientists to explore it in the 1950s. It is also known as the Law of the First Wavefront, which attempts to convey the salient point that the first wavefront to reach either ear will take precedence over the one arriving slightly later in the second ear, and the sound will be perceived to be coming from the direction from which it first arrived. Haas showed that this effect was so strong that it would even beat significantly conflicting Interaural Level Difference cues. In other words, even if the later-arriving signal was significantly louder than the first-arriving signal, the listener would still perceive the sound to be coming from the direction where the sound arrived first. Time-Delay based panning therefore is often significantly more powerful and effective than gain panning.

Modern Attempts (HRTF)

Binaural panning techniques incorporate both Interaural Level Differences and Interaural Time Differences as well as other psychoacoustic effects such as Interaural Spectral Differences in effort to attempt to model human hearing as exactly as possible. Head Related Transfer Functions can be used to create filters that change the frequency response independently in each ear based on the real-world physical size, geometry and materials of human heads and ears. The head can create a shadowing effect on some frequencies at certain azimuths, and the pinna of the ear also creates very complex filtering effects in each ear depending on the source sound positions. These techniques can be extremely convincing under some circumstances and have great application for fields such a Virtual Reality.

Challenges

Unfortunately the circumstances in which HRTFs really work well are very restrictive. Namely these techniques really require headphones for the best result, and even then, HRTFs do not always translate well from one person to another due to quite large variation in human pinna size and shapes. A unique custom set of HTRTFs for each user is the ideal solution but is obviously very impractical. Furthermore, interaural filter effects are extremely sensitive to head position, and indeed this is likely part of their evolutionary design. If humans are forced to keep their head perfectly still their spectral-filtering based localization abilities are decreased significantly. The ear is basically designed to take multiple snap-shots of a source sound from different angles and then effectively triangulate a position based on how the filter effects in the spectrum of the source sound change depending on head angle. If you have ever observed predator animals in nature or those who were trying to avoid them, and wondered why they move their head rapidly in a jerky start-stop manner, you now know why. They are attempting to better locate a respective meal or avoid becoming one. Some form of head tracking and dynamic filter computation would be required to get really convincing results with HRTFs. Again this may find application in Virtual Reality, where head tracking may be incorporated into VR headsets, but it is generally way too cumbersome for professional music applications and simply does not work over loudspeakers.

Both pure Time-Delay based panning and Spectral based techniques also create other potential issues such as concerns over mono-compatibility and introduction of potentially strong coloration of reverb tails if such techniques find their way into the signal sent to a reverb device. Both techniques introduce a form of comb-filtering and if such filtering is altered downstream either by collapse to mono or by recirculating through feedback in an algorithmic reverb the result can be unpleasantly obvious. In general these type of strong comb filtering effects are highly undesirable and often cited by music engineers as a reason for avoiding simple Haas panning and anything based on HRTFs. This is in large part why pure gain panning persists today.

The Solution

Precedence utilizes some variation of all of the above techniques, but does so in a novel manner that largely avoids the issues associated with more advanced time and spectral methods. It uses a proprietary mix of the perceptual cues humans use to locate sounds in 3D space that has be optimized specifically to work within the needs of music and professional audio production. The relative strength of these cues can even be adjusted to suite the needs of a wide variety of musical styles.

Precedence's lateral azimuth positioning ability is extremely convincing over both loudspeakers as well as headphones. It is largely mono-sum compatible even at extreme settings, and if exactly perfect mono-sum compatibility is required it can be guaranteed as needed to avoid bass cancelation in club systems for example. Precedence also offers three input modes to maximize compatibility with the needs of user's audio tracks and sample instrument libraries to avoid this additional potential source of unwanted coloration from spatial information contained within the source sound.

Precedence is designed to be used as an insert directly on individual tracks, and therefore its output will inevitably find its way into reverbs whether placed directly after Precedence on track inserts, or on sends. Unlike other techniques such as HRTFs, Precedence will cause absolutely zero negative consequences when used this way, and the reverb tail will not be undesirably colored as an unwanted side effect. Indeed 2CAudio reverbs such as Breeze 2, have been designed to be used specifically with Precedence in mind. Both are extremely efficient and mid-level recent CPUs can run hundreds or even thousands of instances of both simultaneously to allow users to perfectly spatialize and position every single performer in an symphony orchestra performance simultaneously if desired, all with independant and dynamically modulating settings.