Page 225 - Master Handbook of Acoustics

Eliminating Decay Fluctuations

The measurement of reverberation time by the classic method described above requires the recording

of many decays for each condition. Schroeder published an alternative method by which the
equivalent of the average of a great number of decays can be obtained in a single decay. One
practical, but clumsy, method of accomplishing the required mathematical steps is to:

1. Record the decay of an impulse (noise burst or pistol shot) by the normal method.

2. Play back that decay reversed.
3. Square the voltage of the reversed decay as it grows.
4. Integrate the squared signal with a resistance-capacitance circuit.

5. Record this integrated signal as it grows during the reversed decay. Reverse it and this trace
will be mathematically identical to averaging an infinite number of traditional decays.

Influence of Reverberation on Speech

Consider what happens to the spoken word “back” in a reverberant space. The word starts abruptly
with a “ba …” sound and ends with the consonant “… ck” which is much lower in level. As
measured on a graphic-level recorder, the “ck” sound is nominally about 25 dB below the peak level
of the “ba” sound and reaches a peak nominally about 0.32 second after the “ba” peak.

Both the “ba” and “ck” sounds are transients that grow and decay after the manner of Fig. 11-2.
Sketching these to scale yields Fig. 11-12. The “ba” sound grows to a peak at an arbitrary level of 0
dB at time t = 0, after which it decays according to the RT reverberation time of the room, which is
60
assumed to be 0.5 second. The “ck” consonant sound, peaking 0.32 second later, is 25 dB below the
“ba” sound peak. It too decays at the same rate as the “ba” sound according to the assumed 0.5-
second RT . Under the influence of the 0.5-second reverberation time, the “ck” consonant sound is
60
not masked by the reverberation of “ba.” If the reverberation time is increased to 1.5 seconds, as
shown by the broken lines, the consonant “ck” is completely masked by reverberation. Likewise, a
syllable that ends one word might mask the opening syllable of the next word.