Wave Field Synthesis (WFS)
Wave field synthesis (WFS) is a sound localization technique. It produces synthesized sound waves from a large field of independently controlled loud speakers. The human ear decodes these sound waves as originating from artificial points in space not necessarily associated with the speaker. While stereo or commercially available surround techniques rely on the placement of the listener in a fixed sweet spot for proper localization of reproduced sound, WFS allows for accurate placement of amplified sound sources, regardless of the listener’s relationship to the speakers, while in the active sound field.
In 2015, EMPAC embarked on the design and construction of a high-density wave field loudspeaker array with the primary goal of increasing the high-frequency resolution of current horizontal WFS systems, while creating a modular, easily deployable, and accessible solution for research, creation, and presentation.
In WFS, the distance between the loudspeakers’ centers determines the highest frequency at which synthesized wave fields can physically be rendered. Above this cutoff frequency, distortion of the wave field occurs due to spatial aliasing of higher frequencies. The majority of the systems currently in use have speakers spaced between 10cm and 12cm apart. A WFS system with 4-inch drivers spaced at 12cm will produce an effective wave field with a maximum frequency of approximately 2900Hz; at 24 cm (8-inch drivers), the top frequency is roughly half that, 1400Hz. The localization of sound and the definition of timbre in speech and other acoustic events lean heavily on frequencies 2kHz and above. A motivating question for us was: How do WFS systems that are only reaching around 2900Hz create localization effects and what are the auditory consequences of doubling this range in a large sound field?
Based on these questions, we pursued the concept of an array with smaller distances between loudspeakers. The EMPAC WFS system spaces the loudspeakers at 5.9cm, which results in a system capable of generating coherent concave, convex, or planar wave fronts up to roughly 6000Hz. The EMPAC array has the added benefit of eliminating every other driver from the field, so one is afforded a real-time A-B perceptual study of what these high-frequency components contribute.
The array is comprised of 3 primary systems:
- 16 self-powered speaker modules with 31 independent channels each, adding up to a total of 496 individually addressable channels.
- An uncompressed, low-latency, multi-channel audio network built on the DANTE protocol.
- IRCAM SPAT software suite running on 2 Apple Mac Pros.
(A small ancillary subwoofer array can be deployed to fill out the lower octaves below 140Hz, which are not realistically attainable given the size of the WFS driver and the geometry of its enclosure.)
The speaker array in total includes:
- sixteen 187 cm-long units totaling close to 30 meters of linear amplification when fully assembled.
- Each unit is internally subdivided into 31 loudspeaker enclosures housing a ported 5.5 cm “full range” electrically and mechanically decoupled driver, 5.8 cm on-center from its neighbor.
- These loudspeakers act as individual 20 watt cells (wavelet sources) in the assembly or synthesis of artificial sonic wave fronts to create virtual acoustic environments via wave field synthesis.
A desire for flexibility and exploration led to specific design considerations. The simulation of existing “real” acoustical environments (like the imprint of a concert hall heard in a smaller, significantly different listening environment) may certainly be a worthwhile research goal in order to understand more about acoustics in the natural world and hearing via analysis and re-synthesis. However, from an artistic perspective, this could be seen as rather limiting, since works of art are not always interested in imitating or simulating existing realities as closely as possible.
Building the system out of 16 linear arrays allows the end user to scale and form the system to the size and geometry of a given workspace, as well as experiment with smaller groupings of units placed in and around a performance or listening space in horizontal, vertical, or other orientations.
Each array has the following specifications:
- The ported cabinets are made of MDF and stuffed with polyester fiberfill.
- Mounting anchors and hardware to match on the top, bottom, and rear of the cabinets for the physical positioning or rigging of the modules.
- The rear of each WFS unit holds two 16-channel fan-less Atterotech, purpose-built, power amps.
- Thirty-one of these available channels supply 20 watts RMS to each driver in the enclosure.
- The power amp’s input is derived from a DANTE Network, built on two 24-port IP switches and four DANTE PCIe cards, which are housed in four Thunderbolt expansion chassis nested in an Apple Mac Pro rack.
- The DANTE network allows for a very rapid deployment and reimagining of the physical array, as it requires the engineer to connect just two CAT 5 cables and power to activate a speaker module.
The software used to drive the array is by default IRCAM’s SPAT. This application suite was chosen because of its development in MAX/MSP, its fairly broad user base, maturity, vast external object tool set and capacity to readjust to a myriad of multichannel output states other than WFS, such as Ambisonic. Hybridized playout states can be achieved with this feature. The first iteration of the system runs on two Apple Mac Pros, each handling half of the array. However, the system’s hardware, from the speakers back through the DANTE network, have no fealty to either computer hardware or wave field code, availing itself to more powerful systems as technologies leap forward in processing speed. If a computer will run DANTE drivers and has adequate speed it can mount and run the EMPAC WFS array.
It is EMPAC’s aim to provide a technically approachable wave field synthesis platform that is architecturally open, thus allowing researchers, artists, and engineers space to rapidly begin exploring and creating new works.