from Fig. 3.19, there is not just one test that gives the single answer for speech intelligibility. Rather there are many different results that depend on the details of the test and the type of material presented to the listener. The prediction of speech intelligibility in an enclosed space thus combines the results of listening tests with knowledge of the room's acoustical properties in such a way as to produce a predictable outcome.

Energy Buildup in a Room

When a sound is generated by a single source, the listener receives, in rapid succession, the direct-field signal followed by individual early reflections, and a rising swell of merged reflections whose sum becomes the reverberant field, which finally decays at a rate characteristic of the space. Figure 17.14 shows an example of the idealized pattern. In this figure the three temporal regions are neatly separated; in practice the divisions are not so distinct. The early reflections and the reverberation may be merged. If long-delayed reflections are present they may arrive during the reverberant decay. Sometimes individual reflections are louder than the direct sound when focusing or grazing attenuation is present. The reverberant field can be louder than the direct sound when the receiver is a relatively long way from the source.

The time between the arrival of the direct sound and the first major reflection is called the initial delay gap. If this gap is short enough, early reflections can contribute to increased

Figure 17.14 Idealized Acoustic Response of a Room to an Impulse Excitation

Figure 17.14 Idealized Acoustic Response of a Room to an Impulse Excitation

intelligibility, a broadening of the sound image, and a pleasant augmentation of the sound level. If it is too long, its effect will be to decrease intelligibility.

Background noise, along with long-delayed reflections and persistent reverberation, serve to decrease intelligibility. Background noise that interferes with the comprehension of speech can originate from many sources: people, HVAC systems, exterior noise sources such as traffic, or electronically generated masking noise, which is purposefully introduced to increase speech privacy. When words are spoken in a room, the reflections off the walls and other surfaces will eventually have a negative effect on speech intelligibility, either through long delayed individual reflections or as part of the general buildup of background noise. Thus the reverberant field of speech itself can also become the source of masking noise.

Room Impulse Response

Although it is possible to measure speech intelligibility directly in an existing room, it is also useful to have algorithms to predict it before a room is constructed. As we discussed in Chapt. 11, the impulse response completely defines the properties of a system, and we can predict the result of introducing an arbitrary forcing function (speech) by convolving (integrating) the input with the room's impulse response (Eq. 11.40). An exact formulation of a room's response is not available a priori, but it can be approximated by using the simplifying assumptions or by ray tracing.

A simple model assumes that the room is diffuse and that there exists a reverberant field characterized by a reverberation time. This model ignores common acoustical defects such as long delayed reflections, flutter echo, focusing, and the process of reverberant sound buildup. More complicated analyses utilizing ray tracing can describe these effects, but they are not expressible in a closed-form equation and are time consuming. The approximate methods yield results that are sufficiently accurate, as long as steps are taken to avoid the acoustical defects, which they do not include. The approach is to use the direct and reverberant sound energy densities previously discussed. The direct field (Eq. 2.56) energy density is given by

E WS p2

Sc0t Sco po c0

0 0

Post a comment