Understanding latency with The Liquid Channel

This article discuses the alleged issues of latency when recording, particularly concerning The Liquid Channel.

It is popular belief that even very small delays incurred are problematic and values as high as 3 or 4 milliseconds are can cause problems for musicians. In actual fact, the latency derived from a typical analogue recording arrangement, produces a more unnatural situation than when recording with The Liquid Channel.

The best way to illustrate this is to take three situations whereby a drummer with a snare is listening to what he/she is playing; a 'normal' situation without any extraneous equipment, where the direct sound is travelling from the drum to his or her ears (Fig. 1); one where the drum microphone is connected to an ISA 430 MKII, with headphones attached (Fig. 2); a final situation where The Liquid Channel is being used rather than a MKII (Fig. 3).

In the first example, the distance from the snare to the drummer's ears is taken to be around 2 feet 4 inches, or about 0.7m. The speed of sound, at 340 meters per second, means that the drum will be heard 2.1 milliseconds after it is hit. This is the 'normal' listening situation.

The second instance shows an analogue recording configuration using close miking techniques, where the sound travels almost instantaneously from the microphone to the headphones, producing a latency of only 10 microseconds! Ironically, the sound arrives at the drummer's ears significantly quicker than in a 'normal' listening situation.

The last example illustrates the same situation, recording and monitoring with The Liquid Channel. Here, the latency is on average around 3 milliseconds, less than 0.9 milliseconds longer than a 'normal' listening situation. From these instances, it can be observed that the least natural of the recording configurations is the standard analogue setup, with The Liquid Channel only slightly deviating from the norm. The brain is easily equipped to deal with such temporal shifts and the drummer can continue playing along happily in all cases.

"But what about recording into Pro Tools?" I hear you cry. Well, a Pro Tools system only creates 2 milliseconds of delay (HD @ 44.1 kHz). This means that worst case scenario the delay is still negligible at 5 milliseconds.

A look at studies into the effects of latency on ensemble playing and listening reinforces the point that these less substantial figures have little or no effect on timing but more on the overall perception. When a rock band are playing on stage, the distance of the monitors from each band member can be fairly sizeable, yet they generally have no problem playing in time. Surprising, considering that a distance of just 3 or 4 meters can cause a difference in latency of around 10 milliseconds.

Furthermore, an investigation into such matters documented in a Stanford University paper (http://www-ccrma.stanford.edu/~cc/soundwire/etdea.pdf), found a delay of 11.5 milliseconds to be the 'sweetspot' at which musicians played most accurately at the defined tempo. Above or below that value resulted in instrumentalists slowing down or speeding up respectively in the majority of performances. Only at much higher values were musicians unable to synchronize and substantial timing problems occurred.

Similarly, an audience member attending an orchestral performance will typically be seated around 10 meters further away from the percussion section than the string section, resulting in a delay of approximately 30 milliseconds. This will undoubtedly be somewhat masked by early reflections and the reverberant tail but no noticeable timing discrepancies are heard.

"These facts suggest that latencies responsible for asynchronies in external events of up to at least 30 milliseconds may be considered normal and acceptable under most circumstances; music performance with traditional instruments is not impaired by them".

(ref. http://gsd.ime.usp.br/~lago/masters/latency-paper.pdf)

To answer another specific question, relating to the use of The Liquid Channel in recording the top skin of a snare whilst the bottom is being recorded through a standard analogue configuration; again, this won't produce a problematic time delay. Using traditional recording methods for the top and bottom of the snare (Fig. 4) creates a time delay of 0.5 milliseconds, due to the distance between skins of 5-7 inches. This delay causes a comb filter effect, which creates phase cancellation at around 1k-1k250. Using The Liquid Channel on the top skin means that there is a delay on the top (earlier) signal of (worst case scenario) 3 milliseconds, meaning that the total delay is 2.5 milliseconds. This just means that the comb filter effect is merely shifted to between 150 and 200Hz and still no timing problems arise. If this is still an issue, signals can always be shifted, by creating a delay in Pro Tools for example, or two Liquid Channels could be used which would eliminate the delays altogether.

In conclusion, signal delays of the order of 3 or 4 milliseconds (the time it takes for sound to travel less than a meter) are extremely small and are something that the brain successfully adapts to on a regular basis. Research suggests that latency below 10 milliseconds, or even up to 30 milliseconds in some circumstances, creates little or no problems with regards to cueing, but does have an effect on timbral perception and filter effects.

The Liquid Channel does not create new phenomena or cause timing problems that make performance difficult in a live or studio session, it merely shifts expected phenomena within the limits of our capabilities with no detrimental effects to the instrumentalist. Using traditional close miking techniques creates a somewhat unnatural situation where a drummer hears the sound at virtually the same time as the event. The Liquid Channel, on the other hand, creates a much more natural situation where a drummer hears the sound at a similar delay to when listening acoustically.