Published in August 2004

25 Years of DSP in Sound Reinforcement
By Joseph Bocchiaro III, PhD, CTS-D

Audio enhancement systems are being used in arenas, concert halls and multi-use spaces, such as the Tokyo Internatioanl Forum (left) and the Hamamatsu (Japan) Arena (right).

Digital Signal Processing revolutionized the market.

     With the inclusion of DSP (digital signal processor, or processing) in an ever-widening array of audio products, it is easy to overlook the fact that DSP is now only 25 years old. This makes 2004 an appropriate year to review the state-of-the-art, and to postulate on its future. According to Richard F. Lyon, in his column “DSP 4 You,” (acm queue, March 2004), “…1979 saw the design of the first wave of user-programmable DSP chips: the Intel 2920, the NEC uPD7720 and the Bell Labs (AT&T) DSP-1….The conversion of traditionally analog media to digital began slowly, with the invention of PCM (pulse code modulation) by Alec H. Reeves in 1937. The 1948 paper, ‘The Philosophy of PCM,’ by Barry Oliver, John Pierce and Claude Shannon of Bell Labs, following on their use of PCM for secret wartime communications between Churchill and Roosevelt, laid out an amazing and far-thinking view of why and how digital was going to take over the analog world. The rest, as they say, is history….” Our industry has eagerly adopted this technology, as we will explore here.

Figure 1. Simple DSP audio system.

Evolution of DSP Audio
     As with many technology revolutions, there is a steady parallel development of both the applications and the underlying, enabling equipment and techniques. In the field of sound reinforcement, the requirements to convert sound into a digital signal (A/D), to process it in some fashion and to convert it back to sound (D/A) are very demanding. As the ability to create more complex processors has become possible, the types of applications have evolved from traditional analog circuit replacements to entirely new possibilities.
     At the same time, the audience expectation of the quality of the sound has increased, necessitating higher and higher sampling rates and resolutions. The vast consumer market has fueled the demand for cost-effective DSP chips, and for entirely new programming techniques. As will be seen, DSP has not merely replaced the common analog tools that it emulates, but in many cases makes new techniques possible.
     Speaking at the Richard C. Heyser Memorial Lecture Series of the Audio Engineering Society in 2000, James A. Moorer speculated on some of the trends in his presentation, “Audio In the New Millennium.” He postulates that “extrapolating over the next 20 years, it is concluded that the main problem facing digital audio engineers will not be how to perform a particular manipulation on sound, but how the amount of power that will be available at that time can possibly be controlled.”
     This excellent presentation goes on to explore the possibilities of thousands of channels of digital audio that could be possible because of the processors available, but saddled with the limitations of humans attempting to utilize it. Moorer looks beyond the current surround sound techniques to recording and auditioning spaces that more closely emulate the original sound field.
     What you are reading here coincides with the introduction of the first mass-produced active field control product (see “Active Field Control”), which is a major step toward the multiple-channel, calculation-intensive future of audio DSP that Moorer envisions. The currently available individual DSP techniques are defined, with an eye toward how they are combined in a fully integrated, fully programmable DSP future.
     It is important to remember that these techniques have been developed to create a more realistic sound space, with adequate perceived gain, high intelligibility, low harmonic and temporal distortion, high signal-to-noise ratio and a correspondingly low listener fatigue level. When applied judiciously, each of the following techniques alters the signal in a manner that more closely matches the source with the acoustic space, or alters it in a creative manner that the artist wishes to convey.

DSP Signal Processors
     The use of the term “processors” here refers to a wide variety of equipment packaging that differs between manufacturers, applications and integrations. Some processors are stand-alone, discreet rack-mounted pieces of equipment that require integration within an audio system. Some are multi-purpose devices that allow customized configurations to suit the application (see Figure 1). Some are software products that are companions to computer applications, often referred to as “plug-ins.” Still others are modules available to be installed into consoles, amplifiers, etc.
     Whatever the case, the concepts remain the same, and the designer is left with the challenge of determining the appropriate overall configuration of the system. Many DSP audio product manufacturers offer their equipment in several of these different formats, allowing designers still more options.
     Besides the audio processing aspect of the devices, DSP has opened the door to new and more versatile audio system control possibilities. Because the signals are processed in the digital domain, it is possible to change parameters by programming or “on-the-fly,” remotely or in conjunction with other equipment. This possibility has transformed the live stage show industry in ways equally important to the sound manipulation possibilities.
     In addition, the digitization of the signals allows them to be transported and stored in a pristine fashion, allowing sound engineers the ability to automate performance settings while maintaining unprocessed signals for subsequent use. Control protocols such as MIDI (musical instrument digital interface) have evolved to exploit the capabilities of the DSP products, not just for recording purposes, but for live sound applications.

Signal Delay
     Delay, a relatively simple concept, is now taken for granted because it is incorporated into so many different audio products. One of the most important audio tools, delay is used on a “micro” level to acoustically align transducers in a crossover environment, and on a “macro” level to compensate for distances in large spaces with distributed transducers. The development of easily adjusted delay components has coincided with an understanding of, and the ability to measure, time alignment in a sound reinforcement system.
     It may be argued that this most simplistic contribution of DSP technology has improved the intelligibility of systems more than anything else. In addition, the development of the integrated circuit (IC) “chips” to provide delay has financed and led to many of the more sophisticated DSP applications, such as reverberation and echo.

Figure 2. Feedback suppressor in audio system.

Feedback Suppression
     Feedback suppression products often are thought of as “insurance policies” built into sound reinforcement systems because they have been developed to automatically detect and correct feedback that is occurring in the live environment. These devices have become increasingly more sophisticated in their ability to discriminate when and what constitutes feedback, and have become higher quality components as sampling rates and resolutions have increased.
     In addition, there are now several completely different types of feedback suppression components, allowing designers to select which is the most applicable in each system design (see Figure 2). The devices are useful in cases where the Potential Audio Gain (P.A.G.) of the system is close to the Needed Audio Gain (N.A.G.) of the acoustic environment but not quite attainable. In these cases the systems often are adjusted just below the onset of feedback.
The original feedback suppressors are based on the “parametric equalizer” approach, i.e., they introduce variable “notch” filters into the digital signal chain. The circuitry constantly samples the signal, and in the event that the digital “signature” of feedback is detected, a corresponding filter is activated at the exact frequency. The depth of this filter is adjusted constantly until the feedback is controlled.
     This type of component has been highly refined by reputable manufacturers such as Sabine, Shure, dbx, Peavey, Roland, Behringer, Samson and others. Of course there are limitations to these products, most notably the occasional inability to distinguish certain signals (such as sustained organ notes) from feedback. Because these devices essentially are equalizers, they may “color” the sound as well, due to phase inaccuracies and frequency contouring, often causing undesired results. In a well-adjusted sound system, however, their use as “insurance” against unexpected events is invaluable.
     A second type of feedback suppressor is the “frequency shifting” variety, popularized by the manufacturer Polyfusion. This device constantly shifts the audio frequency of all signals passing through by as much as 6Hz. This shift is not apparent to an audience unless the amplified signal is compared directly to the original sound. Signal regeneration is not possible because the feedback has been frequency shifted from the original and does not build upon itself.
     The newest and most sophisticated feedback suppressor is the Feedback Canceller, manufactured by Wide-Band Solutions. This advanced application of DSP digitally subtracts the feedback it detects on a continuous basis. An advantage of this type of device is that the entire digital signature is being analyzed constantly, making possible other processing, such as noise cancellation.

Figure 3. Simplified electronic acoustic enhancement system.

Noise Suppression,Cancellation
     Manufacturers of DSP-based audio products increasingly are incorporating noise suppression or cancellation capabilities. The bane of any sound reinforcement system, particularly in indoor environments, is a steady-state noise such as from air-handling equipment and ducts, fans, preamplifier hiss, etc. This noise contributes to an overall decrease in the system’s signal-to-noise ratio, decreasing intelligibility and causing audience fatigue.
     The acoustical engineering solutions to these problems usually are quite expensive because modifications to HVAC equipment and architecture often are required. The use of a digital device that samples the background noise and cancels it is extremely useful, particularly in audio- or videoconferencing applications.

Dynamics Processing
     Another important, early application of DSP is dynamics processing. This includes compression, limiting, expansion, gating and combinations of these such as companding. Although this is an area where analog circuitry is very advanced, DSP products are catching up to the adjustability, sound quality and nuances of these tools.
     The ability of DSP products to individually process multiple channels simultaneously makes them cost effective, and makes it possible to process individual signals as never before. Besides stand-alone units, dynamics processing is being incorporated in a wide variety of product types, most notably mixers, preamplifiers and matrix routers. The low cost of DSP has also made it feasible economically to utilize multiple-frequency band processors, such that only the aspect of the signal that requires compression is affected, for example. Intelligence and “look-ahead” modes will continue to become popular to further automate dynamics processing, particularly when sound engineers are not in attendance.

Equalization
     Equalization is an extremely important tool in sound reinforcement systems. Most people in the audiovisual industry are familiar with this process because equalization is nearly as fundamental as amplification in a signal path. Equalization is used for many purposes, but primarily for preferential sound shaping, signal “flattening” to compensate for room variations, and for feedback control due to standing waves or other acoustic anomalies.
     DSP equalization has not merely replaced the traditional analog circuits, however. It has made it possible for sound engineers to select the type of equalizer desired in a system, whether graphic (discreet frequency control) or parametric (variable frequency control), or a combination of the two. Most importantly, advanced DSP equalization promises to overcome the phase shifts present at filter overlap regions, a major cause of distortion in crossover circuits.

Mixing, Switching,Signal Routing
     An essential and foundational component in any audio system is the mixer, switcher or signal router. These components have been transformed gradually from audio/mechanical devices to audio/logic devices to fully digital devices over the past 25 years. The possibilities of DSP application in this area are tremendous.
     The groundbreaking TOA Saori pioneered the concept of an integrated mixer/router/signal processor, and has been followed by a dizzying array of products from other manufacturers. Some of these products include the capacity to incorporate nearly all of the other DSP tools discussed here. The use of these devices has in many cases transformed audio systems into programming-intensive designs as opposed to wiring-intensive designs. This is due to the ability to configure a single component into a multiplicity of applications.

Allows Reconfiguration
     The power of this is to allow designers to create systems that may be reconfig-ured at will, such as with multi-purpose rooms, divisible ballrooms, performance spaces and sports venues. This is accomplished by programming the DSP components to become mix-minus, zoned, bi-amplified, tri-amplified or clustered circuits. It is anticipated that DSP developments will further the trend toward audio systems consisting of a multi-purpose DSP device and audio amplifiers!
     Audio effects include a wide variety of techniques to change the timbral, temporal and dynamic nature of sound, often at the same time and often in conjunction with other effects. In essence, these effects are combinations of the dynamics and equalization effects described earlier. They include echo, reverberation, chorusing, phase shifting, flanging, filter sweeping and others. Although most of these effects are not used for typical sound reinforcement, particularly with speech, they represent some of the most creative and interesting applications of DSP.
     These effects often are used to give a performing artist a signature sound, for example. Reverberation, however, the most common effect, is used routinely in live sound reinforcement, whether for speech only or for music. The quality of reverberation techniques varies greatly, as we will cover later, and is one of the most significant areas of development for DSP designers.

Pitch Correction,Shifting, Harmonization
     One of the newer and most significant DSP developments is in the area of pitch correction and harmonization. Industry leaders such as DigiTech, Antares, TC-Helicon and Eventide offer the ability for vocalists to have their off-pitch notes automatically adjusted back into exact pitch in a real-time fashion. Available as both software and as hardware devices, this technique already has revolutionized audio recording sessions and is making its way into sound reinforcement systems.
     This tool typically is applied in a transparent fashion, but is also used as a special effect for signature sounds. Pitch shifting is applied as a companion effect, often to transpose a singer’s voice when required. Harmonization, a close cousin to pitch shifting, allows layering of the original, corrected signal, with one or several pitch shifted notes above or below it. Harmonization usually is programmed to allow for particular intervals creating the effect of several people singing, often with the characteristics of unique vocal groups such as the Beach Boys or CSNY.

Echo Cancellation,Suppression
     Although echo cancellation (EC) and echo suppression (ES) are not used widely in typical sound reinforcement applications, they are important and sophisticated DSP tools. Developed primarily for audioconferencing and videocon-ferencing systems, EC is invaluable in controlling intelligibility and feedback in conference rooms. Pioneers from the audioconferencing field such as Gentner and ASPI Digital (now ClearOne and Polycom), and videoconferencing companies such as Coherent, VTEL, Picturetel and Polycom have incorporated this essential feature into microphone mixers, audioconference hybrids and videoconference codecs.
     This circuitry analyzes the audio signal from microphones and uses a reference microphone to determine the echo signature of the room. It then subtracts this echo, including early reflections and subsequent reflections, from the output signal. As this technique has developed, features such as real-time accommodation have been added, along with individual microphone channel circuitry. Further developments may include longer cancellation durations, decreased cost, higher sampling resolutions and incorporation into other types of audio devices.

Electronic Architecture
     The most sophisticated application of audio DSP has been developed over the past 20 years, and is utilized primarily in large multi-purpose performance spaces. There are several purposes for this technology, and several variations. The concept of “electronic architecture” or “electronic acoustic enhancement” is applied to spaces where the inherent acoustics of the space are not suitable for the type of performance. For example, opera houses, symphony halls and drama theaters typically are designed with very different reverberant fields and effective degrees of diffusion.
     Electronic architecture systems, utilizing DSP technology, allow the acoustics of the space to be transformed to best suit the situation by emulating natural reflections. This technology is also applied to spaces deliberately designed with little acoustic coloration of their own, with the intention of adding the desired effect to simulate a particular hall, a jungle, the outdoors, etc.
Pioneering work in this field has been accomplished simultaneously by several manufacturers. These techniques and products include the Assisted Resonance System (AR), the Multiple-Channel Reverberation System (MCR), the Multiple Channel Ambiophony System (MCA), the Variable Room Acoustics System (VRA), the Early Reflected Energy System (ERES), the Reverberation on Demand System (RODS), the Acoustic Control System (ACS), the Lexicon Acoustic Reverberation Enhancement System (LARES), the System for Improved Acoustic Performance (SIAP) and the Active Field Control (AFC) System.
     Each of these represents a technique utilizing reverberant field sampling microphones, DSP processing and loudspeaker arrays within the space (see Figure 3, page 50). Control over routing, reverberation, gating and other parameters allows electronic acousticians to “tune” the space as desired.

Conclusion
The use of Audio DSP over the last 25 years has expanded into many aspects of our lives. Cell phones, computers, home theater systems, voice recognition and other applications have followed the development of the equipment used in recording and live events. As manufacturers continue to evolve processing power, we may anticipate that DSP will reach into other aspects of our lives, and further enhance the live event experience.

Joseph Bocchiaro, a principal consultant with Electro-Media Design, Ltd. and manager the EMD Western New York office, is the Chair of ICIA’s ICAT, member of AECT and IACC and participates in Sound & Communications’ Technical Council.