Fig. 3.4 Normalized preference score and interaural correlation magnitude IACC as a function of the horizontal angle of a single reflection for two extreme music motifs A and B. A1=1. Data for 6 sound fields and 13 subjects: ◦; Preference scores for music motif A, x: preference scores for music motif B. •: Values of IACC with music motif A, x: values of IACC with music motif B

3.2Preferred Conditions for Sound Fields with Multiple Reflections

We will now discuss the more general case of sound fields with multiple reflections. Subjective preference obtained by analysis of paired comparisons can be described effectively in terms of four orthogonal properties of the sound field: two monaural temporal factors and two binaural spatial factors [Table 3.1, (Ando, 1983, 1985, 1998)]. These factors are binaural listening level (LL), timing of early reflections ( ti), timing of subsequent reflections (Tsub), and the dissimilarity of the sounds presented to the two ears.

3.2.1 Optimal Listening Level (LL)

The binaural listening level (LL) is the average sound pressure level at the listener’s ears. This is the primary factor that influences listening preferences for sound fields in concert halls. The preferred listening level depends upon the music and the particular passage being performed.

Table 2.1 in Section 2.1.2 lists the two music sources that were used in these listening level experiments. Motif A is the slow and sombre Royal Pavane and Motif B is the fast and playful Sinfonietta. For these two musical extremes, the gross preferred levels obtained by 16 subjects were in the peak ranges of 77–80 dBA, 77–79 dBA for motif A and 79–80 dBA for motif B.

3.2.2 Optimal First Reflection Time ( t1)

The timing of early reflections relative to the direct sound is the second major factor that influences listening preference. An approximation for the most preferred delay time has been expressed in terms of the effective duration of the ACF of the source signal and the total amplitude of reflections (Ando, 1985). This approximation holds when the envelope of the ACF has an exponential decay:

where k and c are constants that depend on the subjective attributes as listed in Table 3.2.

Here the total pressure amplitude of reflection is given by

The relationship of Equation (3.1a) for a single reflection, where A = A1, k = 0.1, and c = 1, becomes

Later, we found that the value of τ e in Equations (3.1b) and (3.3) can be replaced by (τ e)min of the running ACF (Ando et al., 1989; Mouri et al., 2000). The minimum value of τ e in a music piece is generally observed in its most active part, the part with the least redundancy, the sharpest musical contrasts, and the one that usually containing the most “artistic” expressive timing information such as vibrato or accelerando in the musical flow. Echo disturbances, therefore, are easily perceived in musical segments where (τ e)min occurs. Even for a long music composition, musical flow can be divided into short segments, so that the minimal values of (τ e)min of the ACF of the whole musical piece can be taken into consideration. This value is useful for matching musical programs to concert halls, for choosing those music programs that will sound best in a given concert hall. The same is true for preferred reverberation times, as in Equation (3.4).

Methods for controlling the (τ e)min for vocal music performances have been discussed (Taguti and Ando, 1997; Kato and Ando, 2002; Kato et al., 2004). If vibrato is included during singing, for example, we can effectively decrease the (τ e)min of the music to better match the acoustics of the hall.

3.2.3 Optimal Subsequent Reverberation Times (Tsub)

Later reverberations play a significant role in sound field preferences for concert halls. In our experiments, the total amplitude A of late reflections was in the range of 1.1 and 4.1, which covers the usual conditions of a sound field in a room. For flat frequency characteristics of reverberation, the times of later reflections (Tsub)

constitute one of the most important preferred conditions (Ando et al., 1989). The preferred subsequent reverberation time is expressed approximately by a constant multiple of the effective duration of the program material (Ando et al., 1982, 1983):

Considering the fact that the effective duration value of τ e is obtained at the tenth percentile (or −10 dB) delay of the ACF envelope of a source signal, the −60 dB decay time of the ACF envelope corresponds roughly to the “reverberation time” contained in the source signal itself, given by (6τ e). This means that the most preferred reverberation time of a sound field [Equation (3.4)] is about four times the “reverberation time” contained in the source signal itself.

The optimal design of a building must take into account its acoustical role. A lecture and conference room should be designed for speech; an opera house should be designed primarily for vocal music but also orchestral music. For orchestral music, it is recommended that a concert hall be selected from one of two or three types of concert halls according to the effective duration of the music programs that will be performed there. For example, Symphony No. 41 by Mozart, “Le Sacre du Printemps” by Stravinsky and Arnold’s Sinfonietta have short values of (τ e)min. On the other hand, Symphony No. 4 by Brahms and Symphony No. 7 by Buckner are more generally typical of orchestral music. Much longer values of (τ e)min are common for pipe organ music, for example, by Bach. Thus, the most preferred reverberation time for each sound source [Equation (3.4)] can potentially play an important role for the selection of the music program to be performed.

3.2.4 Optimal Magnitude of Interaural Crosscorrelation (IACC)

Binaural similarity or dissimilarity of the two signals arriving at the two ears influences subjective preference. All available data with listeners of normal hearing ability indicate a negative relationship between interaural crosscorrelation magnitude (IACC) and subjective preference (i.e., it has been reconfirmed that listeners prefer somewhat dissimilar signals arriving at their two ears). This relation holds under the condition that the maximum value of the binaural, interaural crosscorrelation function (IACF) is near zero interaural delay, with the sound image directly in front and an equal balance between the sound fields for the two ears. If not, then an image shift of the source may occur. To obtain a small magnitude of IACC in the most effective manner, the directions from which the early reflections arrive at the listener should be kept within a certain range of angles from the median plane centered on ξ = ± 55◦. It is obvious that the sound arriving from the median plane makes the IACC greater. Sound arriving from ξ = ± 90◦ in the horizontal plane is not always advantageous, because the similar “detour” paths around the head to both ears cannot decrease the IACC effectively, particularly for frequency ranges higher than 500 Hz. For example, the most effective angles for the frequency ranges of 1 and 2 kHz are roughly centered on ξ = ± 55◦ and ξ = ± 36◦, respectively (see