A laboratory study was carried out to measure the levels at which various traffic noises start to interfere with the ability to relax and enjoy listening to the spoken word. The intrusive noises represented traffic as heard within dwellings and near roads with a high flow of light vehicles and a varying percentage of heavy vehicles.
The acoustic rating scale units which predicted the subjective responses with least scatter were those based on the Llo concept, the most successful being L~o dB(A). The levels equivalent to those measured "outside" the dwelling correlated somewhat better than those measured "inside".
For units based on peak values alone the "inside" measures performed better than the "outside" measures, and those taking account of low frequency noise factors (e.g., "B" and "D" weighting networks) gave results more in agreement than did those with the "A" weighting.
Of the more recently proposed units for noise assessment PNdB and Lea appeared to be as good as the peak measures, whereas LNp, NNI and TNI were apparently less successful.
In terms of Llo values subjects on average set the intrusion level to 6 dB(A) below the speech level. When the speech level is set to 54 dB(A), a commonly occurring everyday level, the intrusion level judged to be just unacceptable was 48 dB(A). This suggests that an acceptable indoor noise le~,el for listening to speech would be of the order of 45 L:o dB(A).