directivity control of loudspeaker system in low frequency
Transcription
directivity control of loudspeaker system in low frequency
7-07 Directivity Control Of Loudspeaker System In Low Frequency Range DIRECTIVITY CONTROL OF LOUDSPEAKER SYSTEM IN LOW FREQUENCY RANGE Vecky Canisius POEKOEL*, Yoshifumi CHISAKI* and Tsuyoshi USAGAWA* * Human and Environment Informatics, Graduate School of Science and Technology, Kumamoto University, JAPAN 2-39-1 Kurokami, Chuo-Ku, Kumamoto City, 860-8555 Japan email : vecky@hicc.cs.kumamoto-u.ac.jp , chisaki@cs.kumamoto-u.ac.jp, tuie@cs.kumamoto-u.ac.jp ABSTRACT This paper proposes a directivity control method of loudspeaker system in low frequency range to reproduce sound in hemispere, i.e. uni-directional reproduction. Uni-directional sound signage is attractive application, and is able to guide people during evacuations in smoke-filled fire emergencies. Usually, a single loudspeaker has omni-directional characteristics in low frequency. A multiple-loudspeaker system using active noise control (ANC) technique is used to develop unidirectional characteristics in low frequency range. This research investigates a two-loudspeaker system that is configured by a primary loudspeaker faced to 0 degree and a secondary loudspeaker faced to +130 degree or +180 degree as examples. Filtered-x Least Mean Square (FxLMS) adaptation is carried out to configure the digital filter of the system. Experimental results show that the system can control the directivity on target direction at low frequencies. More than 10 dB gain reduction at 200 Hz is obtained for the range of +90 degree around the target direction. Keywords : multiple-loudspeaker system, unidirectional characteristics, low frequency range, FxLMS. 1 INTRODUCTION Generally, a single loudspeaker has omnidirectional characteristics in low frequency ranges. Therefore, a single loudspeaker reproduces sound to all directions in a low frequency range. The unidirectional characteristics of sound reproduction is needed to propagate the sound in one direction as spotlight sound towards the listener, while reducing undesired gain at backside especially in low frequency range, in order to improve the quality of sound or speech at location of the listener. Because the backside gain will be undesired echoes at listener's location due to the reflections. A single loudspeaker is used as the sound reproduction system of emergency system for evacuation either inside building or tunnel to guide the direction for evacuation. For such cases, it is necessary to have the uni-directional sound propagation characteristics. The use of bells on the emergency system can provide information about state of emergencies, but they are not able to give the direction of evacuation. Several studies have been done to develop the emergency evacuation system for maximizing human safety in an emergency situations[1-3]. The combination of a conventional bell sound with certain tones have been developed to improve the quality of emergency information through sound, to be easily perceived as an emergency and dangerous situation[1]. However, in emergencies such as smoke-filled fires, it is necessary to provide sound guidance for emergency evacuation that can lead people to emergency exits. Yokoyama, et al., has developed a system of directional sound guidance to give directions using evacuation voice in event of a fire emergency in a smoke-filled tunnel. The system uses array loudspeakers by implementation timedelay technique with appropriate distance between loudspeakers[2][3]. Kakuhari, et al., has developed point sound source using multiple loudspeakers to reproduce unidirectional sound propagation. This system is very effective in shaping uni-directional propagation characteristics of sound source on the speech frequency range of 500 Hz-4000 Hz[4][5], but it is not designed for low frequency range. A multiple loudspeakers system that is focused to produce a uni-directional propagation behavior, is also proposed. Two loudspeakers are configured back to back, and an adaptive filter using FxLMS algorithm, is used to develop the uni- 257 The Proceedings of The 7th ICTS, Bali, May 15th-16th, 2013 (ISSN: 9772338185001) directional propagation of sound at low frequencies. The proposed system is effective to reduce the backside gain around +130 degree to +230 degree at a frequency range of 100 Hz - 600 Hz. one (S) at +180 degree (180) to reduce undesired gain at around +180 degree direction. This research investigates the proposed multiple loudspeakers system. The system will be used to control the directional of sound propagation of reproduction system, that is focused to uni directional that is around 0 degree, and at the same time to reduce undesired gain around +90 degree to +180 degree directions. 2 MULTIPLE LOUDSPEAKERS SYSTEM The proposed control system consists of two loudspeakers, an adaptive Finite Impulse Response (FIR) filter using FxLMS algorithm, and a delay filter to adjust time-alignment between the desired sound and the controlled sound at control point. A block diagram of the proposed control system is shown in Fig.1. There is a microphone which will be used to adapt the adaptive FIR filter coefficients in adaptation stage, and also to monitor the performance in control stage. Figure 1 also shows two acoustic paths, i.e. a primary acoustic path, P'(z), and a secondary acoustic path, S1 ' z . 2.1 Loudspeakers Configurations Based on the purpose of this study, the primary loudspeaker is directed toward 0 degree for propagating the desired sound, while the secondary loudspeaker is directed toward +130 degree for reducing gain at this direction. This configuration is called P0-S130. Figure 2 shows two kinds of configurations that will be used to control the directivity of sound propagation. As shown in Fig.2, there are two sides of the acoustical field which are left and right sides of vertical axis. For uni-directional control purpose, right side field represents direction of the desired sound propagation, while left side field is the controllable zone. Figure 2(a) shows the P0-S130 configuration, where the primary loudspeaker (P) is faced to 0 degree (0) direction and the secondary loudspeaker (S) is faced to +130 degree (130) direction to control the undesired gain at around +130 degree. Other configuration that is shown in Fig.2(b) is P0-S180 configuration, where the primary loudspeaker (P) faced at 0 degree (0) and secondary 258 Figure 1. Block diagram of multiple two loudspeakers system for uni-directional control of sound propagation 2.2 FIR Filter Setting FIR filter, H(z), that is shown in Fig.1, will be used to produce a controlling sound. This controlling sound will be introduced into the acoustic field through the secondary loudspeaker to the microphone side as the controlled side. Although it is possible to determine the characteristics of FIR filter, H(z), an adaptive method is more convenient and easy to manage the system. In this study, FxLMS which is popular in active noise control (ANC) technique is used. A block diagram of the adaptation process is shown in Fig. 3, there are the primary path P z P ' zS2 z, the secondary path ' S1 zS1 z S2 , a model of secondary path Sz , a system delay L(z), and an adaptive filter H(z). A system delay is installed to compensate the delay (a) P0-S130 Configuration (b) P0-S180 Configuration Figure 2. Two loudspeakers configurations 7-07 Directivity Control Of Loudspeaker System In Low Frequency Range , is for secondary path. A secondary path, Sz installed to implement the filtered input signal, i.e. filtered-X. Estimation of the secondary path Sz , and primary path P(z)P'(z)S2(z) are obtained by impulse response measurement method using time-stretched pulse (TSP) [6]. 3.1 FIR Filter Setting Figure 3 shows block diagram used in adaptation stage. Devices and setting parameters used for both configurations are shown in Table 1, and their snapshots are shown in Fig.4. Table 1. Devices and Parameters Setting The coefficients of adaptive digital filter (ADF), H(z), are adapted using FxLMS algorithm, and white noise is used as input. After the ADF converged, the output of ADF, Y(z) = H(z)X(z), minimized the power of E(z) which is expressed as Eq.(1). This condition means that the primary and secondary sound are in anti-phase at control point. . E z =S2 z {P' z Pz L zS'1 zS1 z H z} X z (1) z, will be The obtained ADF coefficients, H an estimate of ideal coefficients shown in Eq.(2). . Parameters / Equipments Setting/Specification Measurement Environment Anechoic chamber Sampling Frequency 8 kHz Tap Length of adaptive FIR filter 150 taps Tap Length of System Delay 120 taps for P0-S130 1 tap for P0-S180 Loudspeakers BOSE-101VM x 2 Microphone Sony-EMC44B ' H z = − 3 P z P z L z S'1 z S1 z (2) RESULTS OF EXPERIMENT The proposed two-loudspeaker system is tested in anechoic room by using two kinds of configurations, i.e. P0-S130 and P0-S180. The configuration P0-S130 is used to highlight desired sound directional toward 0 degree and to observe the gain controlling at around +130 degree. P0-S180 is used to highlight at 0 degree and to observe the gain controlling at +180 degree. Distances between primary and secondary loudspeakers are 650 mm, and 550 mm for P0-S130 and P0-S180 respectively. Figure 3. Block diagram of single FxLMS filter adaptation 3.2 Obtained Directivity As mentioned earlier, the propagation characteristics of a single loudspeaker at low frequencies are almost omni-directional. Measurement results of a single loudspeaker characteristics using different frequencies is shown in Figure 5, that show almost omni-directional pattern at 100 Hz and 500 Hz and uni-directivity pattern at high frequencies, such as 1000 Hz and above. The directivity characteristics of the proposed multiple loudspeakers control system are shown in Fig.6. Experimental results using P0-S130 configuration are shown in the first column of Fig.6, and using P0-S180 configuration are shown in the second column. Each figure shows the propagation characteristics with and without control at frequencies of 100 Hz, 200 Hz, and 500 Hz. The black color solid lines are directional characteristics under control condition, and the red color dashed lines are obtained by single loudspeaker. According to experimental results of P0-S130 configuration, there are more than 8 dB gain reduction around +100 deg. to +170 deg. at 100 Hz, more than 10 dB around +110 deg. to +190 deg. at 200 Hz, and more than 9 dB around +120 deg. to 170 deg. at 500 Hz. Furthermore, on P0-S180 configuration, there are more than 15 dB gains reduction around +120 deg. to 240 deg. at 100 Hz, more than 9 dB around +130 deg. to 220 deg. 259 The Proceedings of The 7th ICTS, Bali, May 15th-16th, 2013 (ISSN: 9772338185001) P0-S130 and P0-S180, to make uni-directional characteristics to reduce the gain at +130 deg. and +180 deg., respectively. (a) P0-S130 Configuration (b) P0-S180 Configuration Figure 4. Snapshot of loudspeakers configurations Experimental results show that P0-S130 configuration system is able to attenuate 10 dB for 70 deg. range at 200 Hz and P0-S180 one is able to attenuate more than 9 dB for 90 deg. range at 200 Hz. In order to implement actual evacuation guide systems, it is necessary to evaluate the system's performance in reverberant environments such as hallways and tunnels. at 200 Hz, and more than 6 dB around +150 deg. to +220 deg. at 500 Hz. REFERENCE Directivity characteristics that are obtained from those systems show that the P0-S180 configuration reduces gain focus at area around +140 deg. to +220 deg., and P0-S130 configuration focus at area around +90 deg. to 180 deg. There are further reductions by P0-S180 configuration; more than 4 dB at 100 Hz, and more than 6 dB at 200 Hz and 500 Hz at the target of acoustic field of +100 deg to +170 deg. [1] L.C. Broer, S.J. Wijngaar, Directional sound evaluation from smoke-filled tunnels, Proc. First International Symposium Safe and Reliable tunnels innovative European achievements, 2004, 33-41 Figure 5. Directivity pattern of single loudspeaker 4 CONCLUSION This paper investigates the proposed system by using two kinds of loudspeaker configurations, 260 [2] S. Yokoyama, S. Sakamoto, S. Tazawa, Subjective experiment on speech-rate of emergency evacuation announcement in a tunnel, Proceedings Euronoise, 1994. [3] S. Yokoyama, S. Sakamoto, H. Tachibana, S. Tazawa, Study on the Application of Timedelay Technique to Public Address System in a Tunnel, Proc. Inter-noise, 2005. [4] I. Kakuhari, H. Hashimoto, K. Terai, Development of a loudspeaker system with a unidirectional radiation pattern in a speech frequency range, Proc. 106th Convention of Audio Engineering Society, 4867, 1999. [5] I. Kakuhari, H. Hashimoto, K. Terai, Development of a loudspeaker system with a unidirectional radiation pattern in a speech frequency range, Journal of AcousticSociety of Japan, 21, 2002, 369-372. [6] Y. Suzuki, F. Asano, H.Y. Kim, T. Sone, An optimum computer-generated pulse signal suitable for the measurement of very long impulse responses, J. Acoustic Society of America, 2, 1995, 1119-1123. 7-07 Directivity Control Of Loudspeaker System In Low Frequency Range (a) P0-S130 configuration, at 100 Hz (b) P0-S180 configuration, at 100 Hz (c) P0-S130 configuration, at 200 Hz (d) P0-S180 configuration, at 200 Hz (e) P0-S130 configuration, at 500 Hz (f) P0-S180 configuration, at 500 Hz Figure 6. Comparison of propagation directivity patterns with/without control between P0-S130 and P0-S180 configurations at 100 Hz, 200 Hz, and 500 Hz. Red color dashed lines show the directivity of single loudspeaker, i.e. one without control and black color lines show one with control 261 The Proceedings of The 7th ICTS, Bali, May 15th-16th, 2013 (ISSN: 9772338185001) [This page is intentionally left blank] 262