E7 Room Acoustics and Speech Perception

Topic

When a sound source emits sound within an enclosed room, a portion of the resulting acoustic pressure reaches the listener directly, without striking any of the room's surfaces; this component is known as direct sound. The remainder of the acoustic pressure produced by the source strikes the room's surfaces—walls, ceiling, floor, and any other objects present—and bounces off in various directions, creating reflections that also reach the listener, albeit via longer paths and at different times than the direct sound.

Each reflection bouncing off a surface strikes another surface, generating yet another reflection, and so on; eventually, the acoustic pressure originally produced by the source reaches the listener from all directions simultaneously, making it impossible to distinguish which specific surface produced any given reflection. This state, in which the reflected acoustic pressure fills the room evenly in all directions, is known as a diffuse field. The diffuse field is maintained as long as the sound source continues to emit; however, once the source stops, the acoustic pressure within the diffuse field does not vanish immediately. Instead, it continues to bounce, gradually losing energy with each new reflection—a process of gradual decay known as reverberation.

The rate at which reverberation decays depends on how much acoustic pressure the room's surfaces absorb during each reflection rather than reflecting it back: the more absorbent the surfaces, the less energy survives each reflection and the faster the reverberation decays, whereas highly reflective surfaces—such as glass or bare concrete—allow the reverberation to decay much more slowly. The quantity describing this rate of decay is known as *reverberation time*: the time it takes for the reverberation to lose enough energy for its level to drop sixty decibels below the level it held the moment the sound source stopped emitting.

When someone speaks in a room with a long reverberation time, the reverberation of one syllable has not yet finished decaying when the direct sound of the next syllable arrives; consequently, the two overlap in time within the listener's ear. This overlap between the reverberation of one syllable and the direct sound of the next fills the brief silences that normally separate spoken syllables and smooths out the rapid intensity changes that distinguish one syllable from another, thereby reducing speech intelligibility just as background noise would—though in this instance, the interference is generated by the room's own acoustic environment.

To quantify the extent to which a room's reverberation reduces speech intelligibility, this model allows for the simulation of the *speech transmission index*: a measure of how well the rapid intensity variations characterizing speech—which reverberation tends to smooth out—survive after passing through the room. A speech transmission index close to the maximum value indicates that these rapid intensity variations reach the listener almost intact, whereas a low index indicates that reverberation has smoothed them out to such a degree that speech intelligibility is compromised. This same model makes it possible to simulate the optimal reverberation time for various room uses—such as a concert hall, a classroom, or a lecture auditorium—determining a time that is short enough not to compromise speech intelligibility when it matters, yet long enough to enrich the sound of music when the room's function allows for it.

ID:2369