6-channel recording/reproduction system for 3-dimensional auralization of sound fields

Transcription

1 Acoust. Sci. & Tech. 23, 2 (2002) TECHNICAL REPORT 6-channel recording/reproduction system for 3-dimensional auralization of sound fields Sakae Yokoyama 1;*, Kanako Ueno 2;{, Shinichi Sakamoto 2;{ and Hideki Tachibana 2;} 1 Graduate School, University of Tokyo, Komaba 4 6 1, Meguro-ku, Tokyo, Japan 2 Institute of Industrial Science, University of Tokyo, Komaba 4 6 1, Meguro-ku, Tokyo, Japan ( Received 13 August 2001, Accepted for publication 11 October 2001 ) Abstract: In order to simulate three-dimensional sound fields in laboratory experiments, a 6-channel recording/reproduction system has been contrived. To record the sound in a real sound field, six unidirectional microphones combined at every 90 degrees are used. As the reproduction system, six loudspeakers are set in an anechoic room and the recorded signals in each direction are reproduced. The advantages of this system are that the principle is quite simple and the listening area is not strictly limited. In this paper, the principle of the system and the reproduction accuracy are reported. Keywords: 3-dimensional sound field simulation, 6-channel recording/reproduction system, Auralization PACSnumber: Rq, Gx, Lb, Mc, Qp, Yw 1. INTRODUCTION When performing subjective judgment tests on room acoustics and environmental noise problems, it is desirable to reproduce the actual sound fields under consideration with 3-dimensional information. For this purpose, such auralization techniques as the trans-aural system using a head and torso simulator (so called dummy head microphone system) and a 2-channel reproduction system have been designed [1 3]. For the trans-aural system, however, head-related transfer-functions for both ears have to be measured for each listener, and complicated signal processing is needed to cancel the cross-talk components between the 2-channel loudspeaker system and both ears. In addition, the listener s head has to be fixed when applying this method. The 2-channel signals recorded through a dummy head system can be reproduced through headphones, but in this case the accuracy of sound source localization is not sufficient and it is not easy to control the sound pressure in the headphones on the listener s head. Accordingly, the authors have invented a new recording/reproduction system, in which environmental sounds * sakae@iis.u-tokyo.ac.jp { ueno@iis.u-tokyo.ac.jp { sakamo@iis.u-tokyo.ac.jp } tachibana@iis.u-tokyo.ac.jp are recorded through a microphone system consisting of six uni-directional microphones onto a digital audio tape recorder and then reproduced through six loudspeakers in an anechoic room [4 6]. This system (called 6-ch. system, hereafter) is very simple in principle and has the advantage that the constraint on the listening position is relatively easy. In this paper, the principle and reproduction accuracy of the system are presented. 2. OUTLINE OF THE RECORDING/ REPRODUCTION SYSTEM Figure 1 shows the outline of the 6-ch. system schematically. The microphone system consists of six uni-directional microphones combined at every 90 degrees in the horizontal and vertical planes. Figure 2 shows an example of the system which consists of six uni-directional microphones (SONY, C-48) with cardioid directional characteristic (see Fig. 3). In principle, the microphones should be combined as closely as possible. In this case, the distance between the opposite microphones is 135 mm. As seen in Fig. 2, an omni-directional microphone of a sound level meter is located at the center point of the microphone set for the measurement of the absolute sound pressure level in the sound field under measurement. The 6-channel signals recorded onto a digital audio tape recorder (SONY, PC208A) are reproduced through six 97

2 Acoust. Sci. & Tech. 23, 2 (2002) Fig. 2 6-ch. microphone set composed of 6 uni-directional microphones (SONY C-48). vertical planes in the same way as for the microphone set. In this 6-ch. system, any special signal processing is unnecessary in principle, but six spectrum-equalizers (SONY, SRP-E210) are used to correct the frequency characteristics of the total system to be within 1 db between octave bands from 63 to 4 khz. 3. Fig. 1 Outline of the 6-ch. recording/reproduction system. loudspeakers (TANNOY, T12) arranged on a spherical surface of 2 m radius in an anechoic room. The loudspeakers are arranged at every 90 degrees in the horizontal and REPRODUCTION ACCURACY 3.1. Reproducibility of the Direction of Arriving Sounds In order to examine the accuracy of the sound field reproduction by the 6-ch. system, the following physical and subjective studies were performed. For these studies, impulse responses at every 15 degrees around the receiving system in the horizontal plane were measured in the anechoic room using the time stretched pulse (TSP) technique [7,8]. In this measurement, a loudspeaker sound source was located at a point 7 m apart from the receiving Fig. 3 Directivity characteristic of the uni-directional microphone (SONY C-48) measured in octave bands using a pink noise. 98

3 S. YOKOYAMA et al.: 6-CHANNEL RECORDING/REPRODUCTION SYSTEM system. (1) Sound intensity vector in the horizontal plane The 6-channel impulse responses for each direction of the sound source measured through the 6-ch. microphone system were convolved with a pink noise. These synthesized signals were reproduced from the 6-channel loudspeaker system and the sound intensity in each octave band in the horizontal plane was measured through an intensity probe (B&K, 4181) at the center point of the reproduced sound field. In this measurement, the sound intensities in x- and y-directions were measured separately and the intensity vector was obtained from these results. Figure 4 shows the measurement results for the sound intensity vector. In these results, it is seen that each vector points in the true direction in the cases of octave bands from 125 to 1,000 Hz, and the magnitude of the vector is almost uniform in every direction. In the 2 khz and 4 khz bands, however, some discrepancies are seen. This error can be attributed to the fact that the microphones of the receiving system are arranged with finite distance between them as mentioned above. To examine this finite difference error, the 6-channel impulse responses were again measured using a unidirectional microphone by rotating it in 90 degree steps at the measurement point. After the same convolution processing as mentioned above, the intensity vector was measured in the same way as in the former case. The results are shown in Fig. 5, in which it is seen that the reproducibility of the intensity vector has much improved in 1 khz and 2 khz bands, whereas some errors still remain in 4 khz band. This result indicates that the uni-directional microphones should be combined as closely as possible. (2) Subjective judgment of the direction of incident sound Next, the reproducibility of the 6-ch. system was examined by subjective experiment. For this experiment, the 6-channel impulse responses measured in every 30 degrees were convolved with an intermittent pink noise (1 s on-time and 0.5 s off-time, three bursts) and they were reproduced by the 6-channel loudspeaker system. The subject sitting at the listening point was asked to judge the direction of the test sound. In this experiment, two kinds of arrangement of the recording and reproduction systems were examined. They are the plus-figure arrangement in which the four microphones and loudspeakers in the horizontal plane are set as shown in Fig. 6(a) and the Xfigure arrangement in which they are set as shown in Fig. 6(b). As test subjects, seven graduate students with normal hearing ability participated in this experiment. When judging the direction of the sound, the subjects were allowed to turn their heads. The average of the judgment results by all the subjects is shown in Fig. 7, in which the diameter of each circle Fig. 4 Reproducibility of sound intensity vector (The 6-ch. microphone set was used). 99

4 Acoust. Sci. & Tech. 23, 2 (2002) Fig. 5 Reproducibility of sound intensity vector (One uni-directional microphone was rotated in each direction). Fig. 8 Results of the judgment test on the direction of incident sound (X-figure arrangement). Fig. 6 Arrangements of the recording and reproduction system. arrangement for practical arrangement in the laboratory. Therefore, it was decided to adopt the X-figure arrangement as the standard experimental configuration and then the direction judgment test was again performed. In this experiment, the judgment test for the direction in the median plane was added to that in the horizontal plane. As test subjects, ten graduate students participated in this experiment. The experimental results are shown in Fig. 8, in which error judgments are scarcely noticeable. As a result, it can be said that the direction of the incident sound can be correctly judged using the 6-ch. recording/ reproduction system. Fig. 7 Results of the judgment test on the direction of incident sound. indicates the relative number of the response. As seen in these results, it can be said that the direction of the incident sound was correctly judged by the subjects in both recording/reproduction arrangements. In a comparison between the two arrangements, the X-figure arrangement seems to be more convenient than the plus-figure 3.2. Reproducibility of the Sound Fields in Concert Halls As an application of the 6-ch. system to actual sound fields, 6-channel impulse response measurement was performed in the four concert halls shown in Table 1. In this measurement, a dodecahedral omni-directional loudspeaker system was set at the center point of the stage and a TSP signal was generated. Based on the experimental study mentioned above, the X-figure arrangement was also adopted in this case. The impulse response signals measured in the original 100

5 S. YOKOYAMA et al.: 6-CHANNEL RECORDING/REPRODUCTION SYSTEM Table 1 Outline of the concert halls under measurement. Variation Volume (m 3 ) Seats Reverberation time (s) A 21,000 2, B 14,800 1, C 5, D E 19,600 1, sound fields were reproduced in the anechoic room and the impulse responses at the center point of the simulated sound field were measured through an omni-directional microphone and a dummy head system [Head Acoustics, HMM-II]. (1) Echo-diagram and room acoustical indices Figure 9 shows two examples of the comparison of echo-diagram (envelope of the impulse response) between the results measured in the original sound fields and those measured in the simulated sound field in the laboratory. The envelope wave forms were obtained by passing the impulse response signals through a numerical RMS detector with a 1 ms time constant. In these figures, it can be visually judged that the echo-diagrams in the original sound fields are well reproduced in the simulated sound field. In other cases, almost the same resemblance was obtained. As typical room acoustical indices, the values of definition D 50, clarity C 80, center time T S and reverberation time in the two-octave band from 354 Hz to 1.4 khz were calculated from the impulse responses measured in the original sound field and the simulated sound field. Figure 10 shows the correspondence of the values between the original sound fields and the simulated sound field. In every result, the plots lie almost on the 45 degree line; this means that the constructions of impulse response in the original sound fields are well reproduced in the simulated sound field. (2) Study of the listening area by subjective test As the actual listening condition in the simulated sound field, it is desirable that the listener be able to move his/her head to some extent. In order to examine the extent, the following subjective experiment was performed using the binaural recording/reproduction technique. At first, the binaural impulse responses were measured in the simulated sound field using the dummy head microphone system by changing its position from the center point to the point 50 cm from the center in 10 cm steps. For each hall, six pairs between the result for the center point and that for each of other points (including the center point) were made and pair comparison tests were performed. In this experiment, two kinds of test sounds Fig. 9 Reproducibility of echo-diagram in concert halls. Fig. 10 Correspondences of typical room acoustical indices between the results measured in the original sound field and those in the simulated sound field. 101

6 Acoust. Sci. & Tech. 23, 2 (2002) Table 2 Five step categories for the pair comparison test. These two sounds are 1) not different at all. 2) slightly different. 3) moderately different. 4) very different. 5) extremely different. were used: one is the raw impulse response and the other is the sound synthesized by convolving the impulse response with violin music (Gavotte by J. S. Bach) of 6 s duration. These test sounds were presented to the subject through a headphone set and the subject was asked to give his/her impression in the five step categories shown in Table 2. Five graduate students participated in this experiment as the test subjects. In the experimental results, high repeatability was found in every result by each subject and mutual correlation between each subject was high; therefore, the arithmetic average of the category number answered by all the subjects was calculated. The results are shown in Fig. 11, in which it is seen that the subjective difference increases with the increase of the distance from the center point. Assuming that categories 1) and 2) are within the permissible range, 20 cm for the impulse response sound and 30 cm for the music are the permissible distance from the center point. This result indicates that the listener may move his/her head within these areas in the simulated sound field. 4. CONCLUSIONS The principle and the accuracy of sound field reproduction of the newly developed 6-channel recording/ reproduction system have been presented. Although very simple in principle, this system can accurately reproduce the original sound fields with 3-dimensional information and therefore it can be applied to various kinds of psychoacoustic experiments. The authors have been applying this simulation technique to such psycho-acoustic experiments as the study of acoustic environments in public spaces [9], noisiness of road traffic noise [10] and loudness and annoyance assessment of HVAC noises [11,12]. As a similar recording/reproduction system to that introduced here, the authors have designed a method to detect 6- channel impulse responses at the player s position on the stage by rotating a uni-directional microphone and simulate the acoustical conditions on the stage. By means of this simulation technique, the authors have been making investigations on stage acoustics [13]. Although the 6-ch. system has been developed as an auralization tool for the psycho-acoustic experiments as mentioned above, this technique could be expanded to a Fig. 11 Results of the pair comparison test on the difference of hearing impression. general audio recording/reproduction system. In such a case, 6 channels are not necessarily needed and 5 channels (4 in the horizontal plane and 1 overhead) or 4 channels (all in the horizontal plane) would also be effective. When making recordings in actual fields, the receiving system should be as small as possible from a practical viewpoint and the authors are now working on the miniaturization of the microphone system. ACKNOWLEDGEMENTS The authors would like to thank Hikari Mukai, Katsuhiro Yasuda and Masato Ikeda for their assistance in examining the performance of the system. This research was partially supported by Grant-in-Aid for Scientific Research (B , B ) from the Ministry of Education, Science and Culture, Japan. REFERENCES [1] M. R. Schroeder and B. S. Atal, Computer simulation of sound transmission in rooms, IEEE Conv. Rec., pt. 7, 150 (1963). [2] D. H. Cooper and J. L. Bauck, Prospects for transaural recording, J. Audio Eng. Soc., 37, 3 (1989). [3] H. Hamada, N. Ikeshoji, Y. Ogura and T. Miura, Relation 102

7 S. YOKOYAMA et al.: 6-CHANNEL RECORDING/REPRODUCTION SYSTEM between physical characteristics of orthostereophonic system and horizontal plane localization, J. Acoust. Soc. Jpn. (E), 6, 143 (1985). [4] S. Yokoyama, K. Ueno, S. Sakamoto, H. Tachibana and H. Mukai, Sound field simulation using a 6-ch. directional microphone system, Proc. Spring Meet. Acoust. Soc. Jpn., p. 833 (1998). [5] H. Tachibana, K. Ueno, S. Yokoyama, K. Yasuda and M. Ikeda, Principle and application of the 6-channel recording/ reproduction system, Tech. Rep. IEICE, HDA , (2000). [6] K. Yasuda, M. Ikeda, S. Yokoyama, K. Ueno and H. Tachibana, Sound field simulation using 6-channel recording/reproduction system A study on the recieving area for subjective experiment, Proc. Autumn Meet. Acoust. Soc. Jpn., p. 789 (2000). [7] N. Aoshima, Computer-generated pulse signal applied for sound measurement, J. Acoust. Soc. Am., 69, 1484 (1981). [8] Y. Suzuki, F. Asano, H. Y. Kim and T. Sone, An optimum computer-generated pulse signal suitable for the measurement of very long impulse responses, J. Acoust. Soc. Am., 97, 1119 (1995). [9] S. Yokoyama, K. Ueno and H. Tachibana, Experimental study on the effect of noises on conversation in public spaces using 6-channel recording/reproduction system, Proc. WEST- PRAC, VII, 843 (2000). [10] S. Yokoyama, M. Ikeda, K. Ueno and H. Tachibana, Experimental study on the effect of road traffic noise by using a 6-channel recording/reproduction system, Proc. 17th ICA (2001). [11] K. Ueno, M. Senju and H. Tachibana, Subjective effect of HVAC noise on music players, Proc. inter-noise, 2000, 337 (2000). [12] K. Ueno, K. Yasuda and H. Tachibana, A psycho-acoustical experiment on HVAC noise, Proc. 17th ICA (2001). [13] K. Ueno, K. Yasuda, H. Tachibana and T. Ono, Sound field simulation for stage acoustics using 6-channel system, Acoust. Sci. & Tech., 22, 307 (2001). 103