Abstract
A technical profile of RealVerb 5.1, a groundbreaking reverberation plug-in designed for surround sound applications, is presented.  Psychoacoustically important reverberation characteristics are discussed, and used to motivate the design and development of RealVerb 5.1.  RealVerb 5.1 features are reviewed, including control over both spatial and spectral reverberation parameters, and the ability to smoothly morph between simulated acoustic spaces.

1 Introduction

Figure 1: RealVerb User Interface

The spatial character of the soundfield is important to the perception of the sound source and environment.  A conversation heard

inside a reverberant stairwell with acoustic energy arriving from all directions around the listener creates a much different impression when listened to just outside the stairwell through an open door which allows acoustic energy to appear only over a narrow range of arrival directions.

In addition, there are a number of scenarios where both the spatial and spectral character of the acoustic environment is in flux.  Consider, for example, a helicopter flying into a large tunnel, a car speeding out of a parking garage, or people exiting an elevator to a lobby with high ceilings and marble floors.  In some cases, such as a room filling with water or having walls moving closer together, the environment itself is changing.

As a result, in synthesizing reflective environments, postproduction engineers need to control both the spatial and spectral components of the reverberation.  Also needed is the capability to smoothly transition between preset environments.

Several high-end systems (see, for example, [1,2]) synthesize spatial reverberation by convolving the input with measured or computed spatial impulse responses.  While the spatial and spectral character of the environment is faithfully reproduced for the measured source and listener positions, modifications to accommodate a changing environment, or moving source or listener are not available.

Systems such as [3] allow “morphing” between presets; however, they do so by crossfading between the output of parallel reverberators.  The difficulty is that blended reverberator outputs are often very different from the reverberation produced by intermediate values of the physical parameters defining the presets.  For example, consider the case of morphing between an echo at 10 milliseconds and an echo at 100 milliseconds.  Morphing halfway between the presets using the crossfading approach gives two half power echos, whereas the desired output is a single echo at an intermediate time delay.

Here, we present an overview of RealVerb 5.1, a surround sound reverberation plug-in which provides physically accurate and dynamically adjustible control over both spatial and spectral reverberation features.  All RealVerb 5.1 controls are continuously adjustible, allowing RealVerb 5.1 to smoothly morph between simulated acoustic spaces.  In the following, RealVerb 5.1 user controls and signal flow architecture are described.  We begin by reviewing the physical acoustics and psychoacoustics of enclosed spaces.

2 Room Acoustics

Signals radiated from a source in a reflective environment interact with the objects and surfaces in the space to create reflections.  These reflections undergo further interactions, each generating additional reflections.  In an enclosed space, the echo density increases over time to the point that the sound heard by a listener has contributions from such a large number of source reflections that it is Gaussian noise [6].

This notion is illustrated in Fig. 2 and Fig. 3, which show, respectively, a number of paths traveled between a source and listener in an example reverberant space, and the corresponding response at the listener to an impulse radiated from the speaker.  Note that there are roughly three components to the impulse response: the direct path, early reflections, and the late-field reverberation or simply late-field.

Figure 2: Reflective Geometry

The direct path represents the signal arriving directly from the source along a straight-line path. It will appear delayed and attenuated according to the source-listener distance.  The direct path may be specified by its arrival direction, time, and energy.

Early reflections appear after the direct path, and are those arrivals which stand out from the other reflected energy by being separated either in arrival time or arrival energy.  Like the direct path, early reflections are characterized by their arrival direction, time, and amplitude.  In addition to propagation delay and spherical spreading loss, however, early reflections experience filtering as a function of source radiation direction and surface and object interaction.  For instance, sources typically radiate high frequencies stronger out the front, and low frequencies more uniformly in direction.  In this way, a front-wall reflection will have a different spectrum than a reflection from a back wall of the same material.  Many surface materials such as carpet and acoustic tile absorb high frequencies more than low frequencies, and therefore impose a low-pass characteristic on the reflected signal.  On the other hand, materials such as glass and plywood are extremely reflective at high frequencies and modestly absorbing at low frequencies, depending on thickness.

Figure 3: Impulse Response

Finally, late-field reverberation is the noise-like combination of the remaining arrivals.  Because of its statistical nature, the late field is often characterized by its overall equalization and decay rate as a function of frequency.  Late field energy arrival as a function of arrival direction can be reasonably uniform, but is often sensitive to geometric details.  Also sensitive to geometric details is the transition between distinct early reflections and the noise-like late field.  A cluttered space or one with irregularly shaped surfaces will have a room response which quickly transitions to a “diffuse” late field.

Note that if the space is only partially enclosed, the late field will decay quickly since at any given time a portion of the remaining energy exits the space.  In an enclosed space, the materials composition and room size have strong influences on decay rate.  Consider that as the late-field response progresses, the arriving energy has undergone an increasing number of interactions with surfaces and objects in the space, and has traveled through an increasing amount of air.  Accordingly, if the materials present preferentially absorb high-frequency energy, then energy at high frequencies is expected to decay quicker than energy at low frequencies.  It turns out that air modestly absorbs acoustic energy, reducing high frequencies much quicker than low frequencies.  As a result, even in environments dominated by highly reflective materials (a cathedral, say), the high frequencies decay more quickly than do the low frequencies.

3 Psychoacoustics Overview

There are a relatively small number of psychoacoustic parameters which are needed to describe the source and environment.  While the list of parameters is a subject of debate, the ones listed below are commonly used [4]; they have precise definitions in terms of the room impulse response features outlined above.

Figure 4: RealVerb Signal Flow

Direction and distance describe the source position, and are related to the direct path direction and the relative direct path, early reflection, and late-field levels.  Source brilliance and warmth are determined by the early energy equalization.

Intimacy describes the remoteness or intimacy of the space.  It is determined by the delay between the arrival of the direct path and reflected energy; the shorter the delay the more intimate the room.  Whether a room appears muddy or clear is indicated by clarity, the ratio of early energy to late energy: the greater the portion of energy arriving within 80 milliseconds of the direct path, the clearer the source.

Room reverberance describes whether the room is wet or dry, live or dead — the sense that the room prolongs source sounds.  Reverberance has two aspects.  Running reverberance is heard during continuous sounds, whereas late reverberance is heard during breaks in the source signal.  Running reverberance is specified by the early decay time (EDT), six times the time taken for source signals to decay to 10 dB below their initial level.  Late reverberance is expressed as T60, the time needed for source signals to decay 60 dB below their initial level.  Room heavyness and liveliness are specified by the variation in T60 as a function of frequency.

Spaciousness and envelopment depend on the percentage of early energy arriving laterally and the crosscorrelation of late-field energy heard at the left and right ears of a listener.  Large lateral energy portions and small crosscorrelations correspond to spacious, enveloping environments.

Continued on Page 2