Comparison of results of initial study in approximately free-field conditions (solid line) (Ref. 2 ) with the results of a two-room validation study (data points and dashed line): (a) fraction of subjects finding some speech audible, (b) fraction of subjects able to understand at least one word, and (c) speech intelligibility scores, all plotted versus values of the uniformly weighted signal-to-noise ratio . The decibel values and arrows on each plot indicate the various threshold values [(a) and (b)] and speech reception threshold (c).
Plan of loudspeaker locations and descriptions of simulated sound fields: Case A: speech and noise only from the same loudspeaker (No. 1) directly in front of the listener, case B: speech and noise only from two separate loudspeakers (Nos. 1 and 3), and case E: speech from ahead only and noise from all loudspeakers.
SRT values for cases A, B, C, E, and F with configurations illustrated in Fig. 2 and described in Table IV . Arrows indicate the two cases for which significance of the difference is given. Solid filled bars: simple spatial release from masking cases. Hatched bars: cases with added early-arriving reflections.
SRT values for cases L–S, all with speech and noise sounds from all eight loudspeakers. Bars with hatched lines show cases with varied reverberation time. See Table V for descriptions of cases.
Mean SRT values plotted versus the logarithm of the reverberation time of the simulated speech sounds. Dashed line and open symbols: no wall; solid line and filled symbols: speech transmission through a wall.
SRT values for cases A, B, I, H, K, and J with separations of speech and noise in the horizontal plane (hatched bars) and in the vertical plane (solid filled bars).
SRT values for cases D, G, N, T, U, V, W, and X with varied masking noise configurations. Cases D, T, V, U, and G do not include a simulated wall; cases W, X, and N include a simulated wall. (Bars with hatched lines are semi-diffuse cases.)
SIT versus for unmodified speech (i.e., no simulated wall) and diffuse noise.
Construction details of the three walls used in the two-room validation tests, (RC, resilient channels; STC, sound transmission class).
Horizontal and vertical angles of the loudspeakers relative to the listener’s head. Angle 0,0 is straight ahead of the listener’s head.
Descriptions of cases A, B, and E to demonstrate simple spatial release from masking and cases C, D, and F to demonstrate the effects of added early-arriving reflections of speech sounds.
SRT values and descriptions of conditions for 24 test configurations. Column “Expt.” indicates to which of the four experiments each case belonged (each experiment used different subjects). Column “Wall” indicates whether a simulated wall was included or not. Column “Speech” indicates the composition of the speech signal: “Direct,” direct sound only; “ER,” direct sound and early-arriving reflections, “ ,” direct sound early-arriving reflections and reverberant sound with ; “ ,” direct sound early-arriving reflections and reverberant sound with ; “ ,” direct sound early-arriving reflections and reverberant sound with . Column “Noise” indicates the composition of the noise signal: “Front,” from only loudspeaker No. 1 directly in front of subject; “Front side,” from only loudspeaker No. 3; “All,” uncorrelated noise from all eight loudspeakers; “Ceiling,” from only immediately overhead; “Rear side,” from only loudspeaker No. 8; “Front-side diffuse,” predominantly from loudspeaker Nos. 1, 3, and 6; and “Rear-side diffuse,” predominantly from loudspeaker Nos. 6, 8.
Descriptions of the effects of added reverberation: cases P, Q, R, and S without a wall and cases L, M, N, and O with a wall.
SITs in terms of values in decibels. “ER,” direct sound and early-arriving reflections; “ ,” direct sound early-arriving reflections and reverberant sound with a reverberation time.
Article metrics loading...
Full text loading...