FIGURE 1: 3D rendering of a critical listening
room using CATT Acoustics showing the front left/cente/right speakers and the two dipole surrounds.
FIGURE 2: Comparison between two listener/loudspeaker arrangements of a THX dipole surround configuration.
FIGURE 3: Comparison between the modal response for two listener/loudspeaker
arrangements of a THX dipole configuration.
FIGURE 4: Comparison of the speaker boundary interference response for two listener/
loudspeaker arrangements of five equidistant matching speakers.
FIGURE 5: Comparison of the modal response
for two listener/loudspeaker arrangements of five equidistant matching speakers.
FIGURE 6: Comparison between the speaker-boundary interference response for two
listener/loudspeaker positions of a subwoofer.
FIGURE 7: Comparison between the modal response for two listener/loudspeaker positions of a subwoofer.
FIGURE 8: This figure illustrates how the soundfields from a THX setup with dipole surrounds combine over time to energize the room in time intervals 0-5, 5-10, 10-15, and 15-20 ms.
FIGURE 9: This figure illustrates how the soundfields from five matching monopoles combine over time to energize the room in time intervals 0-5, 5-10, 10-15, and 15-20 ms.
The emerging 5-channel digital format with low frequency effects channel, commonly called 5.1, offers an exciting new aural experience. The program can be used to optimize the placement of any number of loudspeakers, so this surround configuration will provide a good example. We can illustrate this by examining the THX home theater surround configuration of a left/center/right (L/C/R) group of front speakers and a pair of dipoles used for the surround channels (fig. 1). The origin is in the left lower front corner, with x positive into the room and y positive going left to right. We introduce a new type of constraint in this optimization. The center channel constraint assures that the center loudspeaker remains on the centerline of the room at a loudspeaker-listener distance equal to that of the left front-listener distance. In this way, all of the arrival times from the front speakers are kept equal. Of course, if this is not desired, the constraint can simply not be applied. In this optimization, the front loudspeakers are allowed to range in X, Y, and also Z. The Z search can be used to determine an appropriate elevation above the floor. This will be useful in some instances, while in others the location of the midrange and tweeter may take precedence for good imaging. These particular dipoles are omnidirectional below 300 Hz, so we will consider them as a point source for the optimization.
Another new constraint relation is added to this optimization to maintain that the dipole surrounds follow the listener’s X coordinate, so that the listener will remain in the null. Thus, as the listener moves forward and backward during the optimization, the X coordinate of the dipoles follows this value. The Y coordinate, or spacing from the sidewalls, is optimized over a limited range, and the Z coordinate can assume any value within its range limits. A comparison of the speaker-boundary interference response and modal response of the best and worst solutions found for the THX configuration in a 5.791 m x 4.267 m x 3.048 m room are shown in figs. 2 and 3. The combined weighted standard deviations for the best and worst solutions were 1.95 and 3.80 dB.
Multichannel Music
In addition to the use of 5.1 in home theater, it is also a music-only or multichannel music format. In addition to the previous dipole surround format, another configuration using five matching loudspeakers is also being used. If the physical constraints of the room permit, all five loudspeakers would be equidistant from the listener. To develop the constraints needed for this optimization, we would make use of the previous center channel constraint, as well as a new rear-channel constraint. The rear loudspeakers can be constrained to the front by the use of mirror planes about the listener. This is a dynamic constraint that follows the listener.
A comparison of the speaker-boundary interference response and modal response of the best and worst solutions found for the multichannel music configuration in a 5.791 m x 4.267 m x 3.048 m room are shown in figs. 4 and 5. The combined weighted standard deviations for the best and worst solutions were 2.04 and 4.06 dB.
Subwoofer
The use of subwoofers is growing in popularity, especially with the surround sound formats. The program can provide optimization of subwoofers via a separate optimization over the 20-80 Hz frequency range (or whatever the operating range is). Once the listening position is determined, you can optimize any number of subwoofers.
Figs. 6 and 7 show the results of an optimization using a single subwoofer in a 10 m x 6 m x 3 m room, working in the 20-80 Hz range. For the speaker-boundary interference spectrum, the variation in the frequency response reduces from a range of about 30 dB in the worst case to around 10 dB for the solution found. The standard deviations for the best and worst solutions were 2.7 and 5.6 dB, respectively.
Comparison Between Dipole and Matching Surrounds
The use of computer room modeling software is increasing. Most of these commercial programs use geometrical acoustics to predict objective parameters, such as SPL, T60, the various energy ratios, and lateral energy fraction. Since they are geometrical models, the validity at low frequencies is limited. In figs. 8 and 9, we compare the build up of the sound pressure level (SPL) over different time ranges. A wide range of coverage maps of this type can be generated to help us visualize these surround sound formats. In addition, the binary impulse responses that are generated by a room-modeling program can be used to auralize virtual environments. Have some fun by downloading a free demo version of the Room Optimizer at www.rpginc.com/products/ roomoptimizer/index.htm and CATT Acoustics at www.rpginc.com/ products/catt/index.htm
Conclusions
A computer program has been developed that allows automated selection of positions for listeners and loudspeakers within listening rooms. The criterion for optimum listener and loudspeaker positions within the room is the minimum standard deviation of the combined short- and long-term spectra. A Performance Index based on the standard deviation function was developed and used to monitor the quality of the short- and long-term spectra. Predictions of the spectra are carried out using an image source model. The optimization is carried out using a standard simplex routine. Some examples have been presented. All cases demonstrate the ability of the program to find the best positions for listeners and loudspeakers within a rectangular room.
