Sound Image Costs and Benefits, Part 2

Practical sound system design for the best preservation of imaging. 4/04/2013 9:38 AM Eastern

Sound Image Costs and Benefits, Part 2

Apr 4, 2013 1:38 PM, By Bob McCarthy

Practical sound system design for the best preservation of imaging.


Figure 1: Plan view (starting from the top) of theater system with center main and L/R sidefills. The red line is time reference, blue is delayed speaker path, and purple is perceived image direction. (A) Frontfills delayed to fictitious stage source. (B) Center main joined to frontfills. If center is late then delay can be added to sync frontfill to center or frontfill can be allowed to lead. (C) Upper and lower level sidefills. Lower has favorable outward aim to create point source. Upper is at outer extreme and is aimed inward to create inverted point source (unfavorable). (D) Underbalcony delays with point-source orientation to center mains. (E) Overbalcony delays with point-source orientation to center mains. See larger image.

In “Sound Image: The costs and benefits”, we established how sound image perception can be characterized in three main categories: horizontal location, vertical location, and perceived range (distance). Our goals are to make audience members perceive the sound as plausibly emanating from the source on stage without localizing to the speakers. Our success is measured by how well we minimize angular distortion in both planes, and if we can decrease the perceived range between the stage source and listener without detection. We will now put into practice the image control techniques previously described.

As discussed in part one, a standard theatrical or concert venue requires a minimum of four sound sources to create a reasonably stable image location for a good number of seats: left-low, center-high, right-low, and most importantly, the stage source: center-low. The more level we get out of the stage source, the more we can expand our area of quality image. The closer we can get our left, center, and right speakers to the stage source, the more we can expect to keep the image tracked to the performer. Additionally, we found that our listeners can detect image errors much better in the horizontal plane than the vertical plane. Therefore, when it comes to tradeoffs, we can get away with moving speakers upward much better than moving outward.

Venue Size and Shape

Are there certain venue shapes that are more conducive to low image distortion than others? Certainly. First of all, size matters. Smaller is better for two reasons: We will have a higher proportion of original source level and time is on our side. In any size venue, the original stage source loses 6dB with each doubling of distance. It is a certainty that the source level will be less in the rear than in the front rows. As venue size increases, they add rows to the rear, but they don’t take away rows in the front. Therefore, the level difference between the early seats and rear seats rises for every row that is added in the back. We can use directional loudspeakers and proper placement to make the sound system achieve equal level from front to back, but as we move back the listeners will certainly hear more speakers and less original source. On the positive side, however, is the fact that as we move back in the hall, the angular offset between the source and speakers will decrease in both planes. Therefore, we can get away with a higher mix of amplified sound.

Now onto the venue shape. Skinny makes things easier. Wide makes it more challenging. The classic shoebox or gently widening fan shape can be very favorable shapes, while wide fans, thrust stages, and in-the-round create the ultimate challenge.

Let’s consider the favorable venue shapes first. The primary reason that skinny is better is that we can minimize the horizontal image distortion. Since the horizontal plane errors are so much more detectable it helps to have a room shape that puts physical limits to the game. Nonetheless, a skinny venue alone can’t do it. We need skinny speaker placement as well. In short, if our left and right sources are placed out at the extreme corners of the hall, we will never fool the people located in the outer seats of the early rows. These seats face a double challenge. They have the highest angular offset and are likely to have very unfavorable level ratios. The angle offset is easy to visualize, but for the level part, we have to remember that our stage source has a directional pattern too. The human voice can be roughly modeled as a 60-degree speaker and in the high-frequency range can be quite directional. Therefore, the near outer seats are likely to have an inward, late, off-axis signal from the stage source competing with an outward, early, on-axis response of a nearby loudspeaker. Ouch!

The more inward we can place the speakers the better off we are in terms of image control. How far inward we can place the speakers will be governed by the stage width. Therefore, the extent to which the stage width is narrower than the seating is movement in our favor. The challenging near-outer seats have a fighting chance of imaging toward the stage if the speakers are placed inward to them, thereby minimizing the horizontal image offset. As the stage width goes down proportional to the room width, and our left and right positions move inward, we can tolerate wider halls since the horizontal geometry lines up similarly for both the stage source and speakers. This has it limits, however, since stage sources are not necessarily omnidirectional. This leaves the outer seats with a late, weak, off-axis signal from the stage that is unlikely to pull its weight in the imaging equation.


Sound Image Costs and Benefits, Part 2

Apr 4, 2013 1:38 PM, By Bob McCarthy

Practical sound system design for the best preservation of imaging.


Figure 2: Section view of same theater system with center main and L/R sidefills. Colors same as Figure 1. (A) Frontfills delayed to fictitious stage source. (B) Center cain joined to frontfills. If center is late, then delay can be added to sync frontfill to center or frontfill allowed to lead to bring image down. (C) Upper and lower level sidefills. Neither has favorable location to bring image to stage. Lower may be helped by stage source arrivals. Upper sidefill has favorable orientation to image to center cluster but not to stage floor. (D) Under and overbalcony delays sync’ed to center mains. Since mains are high, only the overbalcony has favorable orientation toward stage. See larger image.

Main speaker placement

In the most basic sense we can model the source and loudspeaker(s) arrangement as a speaker array. The most successful configuration would be the uncoupled point source in both the horizontal and vertical planes. In short, all of the sources reconcile back to a virtual source point behind the original stage source. For example, outward-facing left and right speakers relate as a point source to the stage (favorable) and inward-facing left and right speakers (unfavorable) create an inverted point source, known as a “point destination” array. If the left and right speakers move to a straight-ahead orientation, the array relationship is an uncoupled line source (neutral). Central listeners will be pulled outward away from the stage and those on the sides will find the speakers guiding them inward toward the stage. The center speaker has a greater challenge since we tend to place both performers and audience members on the floor and center speakers high in the air. In the horizontal plane the center speaker lines up very favorably with the source, but in the vertical plane, we are sure to have listeners for whom the center speaker will appear as an inverted point source with tendencies to pull the image skyward. The only way those listeners will have an uncoupled point source orientation to the stage source and center speaker is if they are in a high balcony area looking down on both the stage and the loudspeaker. This is certain to be a minority of the listeners. Fortunately, we are far less precise in our vertical localization than our horizontal, so these errors are less noticeable to the listeners.

So we now have a basic placement strategy for the main system design: left and right should be as close in to the stage as practical (within sightlines and gain before feedback limits), at the lowest practical height that gives us even level from front to back. Center should be as low as possible (almost always a scenic/sight line limit).

Fill speaker placement

The main system can get coverage help from the auxiliary fill systems such as frontfills, sidefills, infills, and delays. Let’s look at the best way to position these systems to aid our image placement goals. First up is the frontfill array. This is the easiest of all possible subsystems to place in an image enhancing location. By being spread across the stage front they are assured of bringing the focus onto the stage horizontally and only slightly below the stage source vertically. The worst-case scenario is we appear to have people singing out of their feet, which is far less an issue than our mains faced. The frontfills have the ideal imaging location but can only be used for a limited distance, typically about three rows.

Next in line is a sidefill speaker, which could horizontally extend the L/R main coverage to the near outside seating. Like the L/R mains, we want to place them as far inward as possible and aim them outward to continue our point source orientation to the stage (and the mains). Keeping these at the lowest possible level (relative to the mains and stage source) will help keep our focus inward.

Infill speakers can extend the horizontal range of the L/R mains to cover the central area beyond the reach of the frontfill. This is a dangerous pursuit as this area can become very crowded with arrivals from the center main, the frontfills, and the infill from the opposite side. The infills are an inverted point source to the stage source in the sensitive horizontal plane, so the image distortion risks are very high. The best strategy for infill arrays is a minimalist one: Keep the level as low as possible so that only a small number of seats are covered exclusively by the infills by giving way to frontfills and center mains as much as possible. In summary, infill speakers should be placed as far onstage as possible and run at the lowest level you can get away with to fill in the gaps.

Delayed speakers present the next challenge. Like the frontfill, under (or over) balcony delays have a very favorable horizontal orientation to the source, once again an uncoupled point source. But like the center main, they are vertically challenged because they are invariably an inverted point source to the stage with an upward orientation. On the positive side, however, is the fact that in many cases, the vertical angle between the stage source and under-balacony speaker is very small, leaving us with a low risk of substantial error. Therefore, we are working to overcome small angular distortions in our less-sensitive plane (the vertical). Contrast this to the far more difficult task faced by the center main, which has extremely high vertical angle distortion in the front parts of the room. The placement of under-balcony delays is described by the component spacing and the depth of the coverage. It is a given that they will be mounted on the ceiling, so we won’t have the option of going lower. In the horizontal plane we are better off not to attempt to use widely spaced, wide-coverage angle speakers. To do so would place many listeners in locations where they may have a substantial outward horizontal image pull toward the delay speaker. Instead, an appropriate quantity and spacing of speakers with 60 to 100 degrees of coverage reduces the horizontal image risks. The vertical image is kept in check by reducing the coverage depth to the minimum required, and therefore allowing us to keep the level low. If the delays have to throw too far they will inevitably be too loud at the rows closest to them, which is exactly where the vertical image distortion risks are the highest. It helps to aim the delays to the deepest area of coverage, thereby keeping the nearest rows underneath their coverage. Far too often we see under-balcony speakers hung out on the front edge of a long balcony and aimed downward when their best placement would be much deeper underneath and aimed at the rear row.

Overbalcony speakers are a variation on the same theme as underbalcony speakers, but with a few twists. First is the fact that we are not necessarily forced to mount our speakers directly onto the ceiling but may have the option of dropping them down until sightline limits apply. A favorably low position can greatly reduce the vertical image distortion risks. In the horizontal plane the same issues apply, but it is worth restating the disadvantages of using a small quantity of overly wide delays up there. It is not uncommon to see designs with a single point source cluster of two or three speakers creating a wide horizontal spread to cover the overbalcony. This is inadvisable since it opens up large angular distortions in the sensitive horizontal plane for those with seats in the sides. If two or three speakers are needed to provide sufficient coverage, it is better to spread them apart in an arc or line and minimize the angular distortion for the center and sides alike. That covers the mains and typical subsystem placements. Now let’s move on to how to set the timing sequence.


Sound Image Costs and Benefits, Part 2

Apr 4, 2013 1:38 PM, By Bob McCarthy

Practical sound system design for the best preservation of imaging.


Figure 3: Plan view of theater system with L/R mains and centerfill. Colors as per Figure 1. (A) Frontfills delayed to ficitious stage source (B) Left main outer delayed to fictitious source. Using the outer location has less image distortion than using the inner location (shown as dotted lines). (C) Center can be delayed to meet L/R system. If center is late, then delay can be added to L/R mains to sync them to centerfill. (D) Underbalcony delays with mostly favorable orientation to mains. See larger image.

Timing Sequence

Since we want things to image to the stage source it will be a good idea to find out how long it takes to get from the stage to various locations in the room. The point of temporal origin is found by having a loudspeaker stand-in for the actor/musician. The placement can be somewhat subjective in the case of an actor, unless they spend the entire show in the same spot. In the case of a drum kit, unless otherwise specified, we can assume it is staying put. This temporary source speaker is termed the “fictitious source.” In the case of a moving actor, the placement should be approximately mid-depth of the actor(s) front/back range. If the source is too close, the delay timing sequence will start too low and the image control will be needlessly compromised. If the source is placed too far upstage, then the stage source will lead by too much time and a new set of problems such as intelligibility loss arise.

Now that the fictitious source is in place let’s recheck the signal path to make sure that there is no digital latency, since current day actors are still running analog. If a digital console, signal processor, amp, or all of the above is driving the source loudspeaker, then there is a latency error that will cause you to read a lower delay setting than that required to accurately sync to an analog actor. Either go analog or measure the digital path and factor the latency into your calculations. From here the procedure is straightforward: measure the time it takes to arrive from the fictitious source, then measure the time it takes to arrive from the local speaker, subtract the difference, and put that time into the local speaker’s signal path. Done.

If that sounds too simple, it is. There are a few details to consider such as where to locate the mic to make the timing calculations and exactly who do we delay to whom. Let’s cover mic placement first: In general we place the mic in the middle of the particular speaker’s coverage, the middle horizontally and the middle depth. This way the timing and level errors accumulate most evenly in all directions from the central zone of coverage.

Now on to who do we delay to whom. This is more challenging. It is tempting to try delaying every speaker in the building back to the fictitious stage source. This would seem to ensure the best timing sequence for minimal image distortion. This is nice in theory but rarely in practice. For this to work in practice we will need to have very strong level from the live stage sources to give us enough level to minimally reinforce at the far reaches of the room. This could be the case for a very small room or with a very loud stage source, in which case the sound system is providing very light reinforcement. In practice we will usually find that the stage sound provides strong enough levels only in the closer areas of the room and most likely not in remote areas covered by our fill systems such as the extreme sides, under, and overbalcony areas. Think about it: If our sound system is having trouble reaching these areas, how well do you think our stage source is doing there? While it may seem desirable to sync these remote areas to the stage source, to do so will degrade the intelligibility in those areas. Why? Because you are in sync to an inaudible source (the stage) and out of sync to an audible one (the mains). In practice, the timing cannot be initiated until we have discerned the level relationships between the stage sources, and each main and subsystem to each other.

The easiest of all to begin with is the frontfill. Let’s begin with the left and right main. In their coverage area, there are likely to be arrivals from the stage, the frontfills, and the center. The center is late and high. The frontfill and stage source come from the same place. Who do we sync to? The stage is fine or the frontfill; they will be about the same. How about the frontfill? Listeners there hear the stage and a very late center and left or right. This is a no-brainer: the stage. The center speaker is the ultimate engineering challenge because we would really like to insert an accelerator on it rather than a delay. It is high and late, and it has digital processing on it. Typical desired setting is -15 milliseconds to -30 milliseconds (negative delay). Typical actual setting: 0 milliseconds. The extreme sides covered by our sidefills will hear an extremely low level stage source and the off axis edge of the left or right main (which was already delayed to the stage). The dominant source here will be the L/R main. If you get tricky and sync to the stage source, you are syncing to a ghost but will have a very real fight with the main speaker in your neighborhood. This is not advised; sync to the main. For the infill, it is again advised to sync to the sidelobe of your (already delayed) main neighbor and not to the stage source. Your inward angle means that you will arrive later than the stage source in your coverage area, which is to everyone’s advantage.

Finally, on to the delay speakers. It is tempting again to want to join the stage source, but once again you are chasing ghosts. The main systems (provided we can see them) are the dominant players under the balcony. If you are not in time with them, it is the same as adding reflections under/over the balcony.


Sound Image Costs and Benefits, Part 2

Apr 4, 2013 1:38 PM, By Bob McCarthy

Practical sound system design for the best preservation of imaging.


Figure 4: Section view of same theater system with L/R mains and centerfill. Colors same as Figure 1. (A) Frontfills delayed to fictitious stage source. (B) Lower main delayed to fictitious source. Upper main is then delayed to lower main at the balcony front. (C) Center can be delayed to meet L/R system. If center is late, then delay can be added to L/R mains to sync them to centerfill. (D) Underbalcony delays with favorable orientation to mains. See larger image.

Setting the Range

The final aspect of the image control scenarios is the range setting. This is done by equalizing the mains and subsystems to have a similar frequency response throughout the room, thereby minimizing the perceived range disparities. As we move to the most distant parts of the hall we can allow the response to tilt somewhat in favor of the lows, thereby letting the far seats move back enough to keep their range plausible.

Out of Time

In this two-part article we covered how we perceive sound image and how we can design sound systems to take practical advantage of speaker placement and timing for the best preservation of imaging. I hope this will be helpful to your designs and tunings.

Events/Training more →
Past Issues
January 2015

February 2015

December 2014

November 2014

October 2014

August 2014

July 2014

June 2014