Natural room reflections are too weak and early to please listeners of reproduced speech.

Evidence

Figure 7.1 shows levels for a single delayed sound, from a horizontal angle of 36°, that listeners reported as enhancing the sound of speech. The early reflections in typical real rooms (the black dots) are all lower in level and earlier in time than the levels preferred by listeners. Multichannel audio is needed to provide added, stronger (by 7 to 10 dB) and later (by up to 30 ms), reflections for our listening pleasure (Ando, 1977, 1998). Note that the preferred levels just avoid the “second image” curve, indicating that the preferred 191. reflections were all within the precedence effect fusion zone, thereby not generating distracting second images. The inevitable conclusion is that, in listening to live speech, and for a single loudspeaker reproducing speech, individual room reflections are not problems. In fact, they are not loud enough or late enough to put the speech into a pleasing acoustical setting. — Toole, Floyd E, Sound reproduction- the acoustics and…, 686

Categorization

• XREF: Reinforces the earlier precedence-effect material; the fusion zone concept links to how comb filtering and echo perception are managed in loudspeaker design.
• SHARE: Good for someone doing acoustic design of listening rooms — the quantitative gap (7-10 dB, 30 ms) between what rooms naturally give and what listeners prefer is a specific, usable figure.