Each weighting factor given in Table 1 represents the fraction of overall speech intelligence contained within the associated one-third octave frequency band.

The weighting factors in Table 1 are obtained by multiplying each individual one-third octave band weighting factor of ANSI S3.5-1969 by 300. Articulation class (AC) values are thus related to but distinctly different from articulation index (AI) values. In particular, the AC considers only the effect of signal attenuation; while the AI considers such additional factors as speech level and spectrum and background sound level and spectrum.

Note 2—The AC is similar to the DAI rating proposed by Warnock⁶ and has been shown to correlate with AI values derived from ANSI S3.5, except where the AI approaches 1 or 0 (AI values range between 1 and 0 and approach 0 with increasing privacy and nonintelligibility). Articulation class values give the reverse. They usually exceed 100 and increase with increasing privacy and nonintelligibility. Extensive comparison between AC ratings and subjective judgments of open-plan speech privacy has not yet been accomplished.

1.1 This classification provides a single figure rating that can be used for comparing building systems and subsystems for speech privacy purposes. The rating is designed to correlate with transmitted speech intelligence between office spaces.

1.2 Excluded from the scope of this classification are applications involving female speakers or children, languages other than English, and sound spectra other than speech. Thus excluded, for example, would be comparisons of building systems or subsystems for their effectiveness in reducing transmitted noise from machinery, industrial processes, bowling alleys, music rooms, places of entertainment, and the like. Published work by Pearsons, et al, may eventually permit the restriction on female speakers to be relaxed.