Enhanced Directional Sensitivity using Acoustic Dish Reflector

Transcription

1 Enhanced Directional Sensitivity using Acoustic Dish Reflector Omkar Sawant, Anirban Bhowmick, Thippur V. Sreenivas Dept. of Electrical Communication Engg. Indian Institute of Science, Bangalore, , India Abstract An acoustic reflector is a passive device which is used with microphones to reflect and focus sound waves from a longer distance. They are effective for long-range sound recording without intrusion, such as to record wildlife sounds. In this paper, our objective is to investigate the feasibility of using a dish reflector microphone for indoor activity monitoring of large premises. Acoustic gain and frequency response of the microphone with different parabolic, near parabolic and nonparabolic reflectors are compared with that of the same microphone without the reflector. It is observed that the use of dish reflector does significantly enhance the directivity and sensitivity of sound recording. Keeping to a moderate size reflector, among various shapes and materials, the best reflector has given a 14 db gain improvement over microphone without reflector at 0 degree azimuth. We also found that among steel and plastic dish reflector of approximately same size and shape, steel dish reflector does give a higher performance as expected. The increased directivity of the dish reflector is giving a significant improvement in indoor monitoring and also for sound source tracking. I. INTRODUCTION An efficient sound signal acquisition is an important problem as much as sound analysis. It is a topic of paramount importance because an improved sound acquisition can reduce the burden of sound processing; however, more effort is needed to improve the quality of acoustic signal acquisition. In the textbooks on acoustics, we do get some references to acoustic reflector microphones; however, the physics behind the phenomenon of concentrating sound waves at a single point has not been exploited enough in practice. J.W.S. Rayleigh [1] brought out this phenomenon by citing examples like the whispering gallery in St. Paul s Cathedral in London and the ear of Dionysus in Syracuse, Sicily. The use of parabolic shells in churches has been highlighted in Cremer s work [2]. In an interesting study, Yoshifumi et al. [3] have proposed a ceiling dome microphone for monitoring breathing sound under room environment. The experimental results proved the effectiveness of the method and the feasibility of its use at home. In an outdoor scenario, Aumann et al. [4] have described the use of small parabolic reflector microphone to capture the buzz of honeybees to monitor the health of beehives. W. Swenson [5] has given a comparison of different reflector systems for field recording of natural sounds and concluded that the reflector microphone is an inexpensive and effective alternative to line array microphones, However the physical environment is also a factor in microphone performance. Dai et al. [6] have /18/$31.00 c 2018 IEEE presented literature on the use of the parabolic reflector to amplify in air signals generated during impact-echo testing. They suggested that parabolic reflector would be effective for air-coupled sensing in both impact-echo and surface wave tests. A comparative study of several theoretical and practical aspects of a conventional parabolic dish microphone have been produced by Chen [7]. Dish reflectors have seen most of its utilization in outdoor applications. They are used to pick up feeble long distance sounds or to record wildlife sounds. It is also used in football games to record audio. In this paper, we have investigated the feasibility of using a dish reflector microphone for indoor activity monitoring of large premises. We have used different parabolic, near-parabolic and non-parabolic shape reflectors with the microphone to evaluate their directivity, sensitivity and noise rejection capability in the indoor recording. Reflectors made of steel and plastic material are compared in terms of sensitivity and directivity. We measured different dish reflectors of various shapes and sizes but finally, we have confined the experiment to moderate size reflectors for the experiment. II. ACOUSTIC REFLECTOR MICROPHONE The nature of wave physics related to acoustic waves and EM waves are similar and therefore the gain of the reflector dish can be calculated using the well-known expression [7] of the dish antenna. The gain of the reflector is the measure of how much input power is received in a particular direction compared to an omni-directional microphone. The expression for parabolic reflector dish gain is: G = η 4πA λ 2 (1) where, A is the projected area of the dish reflector, η is the dish efficiency which is dependent on the construction and material of the dish and λ is the wavelength of the incident sound signal. So, for a dish diameter D, the projected area is A = πd2 4. Now the dish gain can be expressed as: ( ) 2 πd G = η (2) λ Practically, the dish efficiency η is in the range of 45% to 70% depending on the material, shape and construction [7]. Here, we have calculated the gain by assuming efficiency as 50%. At

2 30 the sound velocity is m/s. Thus, we can compute the gain of the dish as a function of frequency and dish diameter. Fig. 1 shows these theoretical gain and as we can observe that (a) (c) (d) (e) (f) Fig. 2. Profile of the reflectors used in the experiment are shown, red line depicts the theoretical parabola corresponding to the dimensions of reflector used whereas the blue line depicts the actual shape of the reflector used (a) plastic dish 1 (dimension D=21.5cm,l=9.6cm) plastic dish 2 (dimension D=16.2cm,l=6.6cm) (c) plastic dish 3 (dimension D=26cm,l=6.3cm) (d) steel dish 1 (dimension D=21cm,l=8.5cm) (e) steel dish 2 (dimension D=24.2cm,l=7.4cm)(f) steel dish 3 (dimension D=24cm,l=6.6cm) Fig. 1. Theoretical gain characteristics of parabolic dish (30 Temperature) it is inversely proportional to the λ 2, which means that the dish becomes more effective at higher frequencies. The parabolic dish diameters depicted in Fig. 1 are the same as those used in our experiment. The focal length is another critical parameter to get maximum amplification of the acoustic signal. The expression for focal length (a) of the parabolic dish is: a = D2 16l where, l is the depth of the dish, D its diameter. III. EXPERIMENTS In order to quantify the performance of the reflector dishes, we conducted experiments inside the anechoic chamber as well as in a sound treated lab. We have used the following reflector dishes having the characteristics as mentioned in Fig. 2. Fig. 3 shows the plastic and steel dishes used as reflectors. From Fig. 2, we can observe that the profile of the plastic dish-1, plastic dish-2 and steel dish-2 have more similarities to its theoretical parabolic shape; whereas the plastic dish-3 has a flat bottom. dish-1 and steel dish-3 have near parabolic shape. In the anechoic chamber, we have set up a measurement procedure where the microphone is placed at the center of a hypothetical circle having a radius of 2 meters and a loudspeaker is placed around the perimeter of the circle at an interval of 20, as well as at 0 (exact front) and 180 (exact back). We have recorded a speech signal for 10 sec in the anechoic chamber and used it as the fixed stimulus. For conducting the experiment, we have connected the reflector microphone to a laptop via Audiobox A/D converter (M-track) with a USB interface. We play the speech file through a loudspeaker and (3) (a) (c) (d) (e) (f) Fig. 3. Reflector dishes used in the experiment (a) plastic dish 1 plastic dish 2 (c) plastic dish 3 (d) steel dish 1 (e) steel dish 2 (f) steel dish 3 the received signal at the microphone is recorded using of Audacity software. We have also recorded a 10 sec silence period after the 10 sec of speech stimuli in each recording session. Fig. 4 shows the recorded speech and silence period in one session using steel dish-2 as a reflector. Figure 5 shows the experimental setup inside the anechoic chamber. Fig. 4. Recorded speech and silence in one session

3 TABLE I EXPERIMENTAL FOCAL POINT COMPARISON; STARRED DO SHOW SOME DIFFERENCE Fig. 5. Experimental setup in anechoic chamber A. Frequency Response Measurement To measure the frequency response of the dish microphone, we have generated sinusoidal test signals of different frequencies using MATLAB and played through the loudspeaker placed at 2 meters distance and 0 angle from the reflector. We have estimated the received signal power and noise power as follows: S.no Dish Type Dish 1 Dish 2 Dish 3 Dish 1 Dish 2 Dish 3 Theoritical Point (cm) Exmerimented Points (cm) Signal power (db) Chosen Point (cm) P sn = P n = N 1 n=0 N 1 n=0 x sn 2 (n) N x n 2 (n) N (4) (5) P s = 10 log 10 (P sn P n ) (6) where, P sn is the received power of signal mixed with recording noise, P n is only recording noise in the absence of transmitted signal and P s is the estimated signal power. We have considered here the recording noise is additive in nature. B. Optimum Aggregation Point of Reflectors We have conducted this experiment to ascertain the aggregation (focal) point of different reflector dishes. We have shifted the microphone position ahead and behind the theoretical focal point for each reflector and then recorded the input stimulus at 0 and at a distance of 2 meters from the reflector; the received signal power are tabulated in Table I. Although, we have only mentioned three adjustment points per reflector in Table I, we have experimentally checked several possible focus points. We note that the experimental and theoretical focal point for the reflectors nearly resembling a parabola are the same; however, in the case of steel dish-1 and steel dish-3, the point of maximum signal power(db) is different from the theoretical focal point. We have explored this property in anechoic chamber measurement by adjusting the microphone position to achieve maximum signal power(db). For nonparabolic plastic dish-3, we can observe that shift between theoretical and experimental focal point is maximum. We can note that the steel dish-2 is giving the most gain and also matching with the theoretical focal point. C. Directivity and Sensitivity Measurement We have performed this experiment to observe the directivity pattern of different reflectors in terms of microphone sensitivity. We have made a comparative analysis of different reflectors depending upon the dish materials i.e. plastic and steel. Fig. 6(a) shows a comparison of all the steel dish reflectors, where steel dish-2 is outstanding amongst the three with a gain of 14 db over the microphone without a reflector. Similarly, Fig. 6 shows a comparison of the plastic dish reflectors. dish-1 is showing the best result with a gain of 9 db over the microphone without a reflector at 0 angle. It also shows that the plastic dish-3 is giving a pencil like highly directive pattern. Fig. 6(c) shows the comparison of the best of the two categories of reflectors i.e. plastic and steel dishes. From the data, we can make an optimum decision of the reflector dish depending on the application envisaged. For capturing wildlife sounds highly directive plastic dish-3 would be most effective, whereas for covering large premises steel dish-2 would be the best contender because of its higher sensitivity. In scenarios where we need high back-lobe rejection, we would favour plastic dish-1. D. Frequency Response of Different Reflectors The frequency response of each of the reflectors is shown in Fig. 7. We can observe that, compared to no reflector microphone, the reflector dishes have maximum amplification in the frequency range of 2-4 khz. Fig. 7(a) shows a comparison of steel dish reflectors with respect to no reflector microphone. It is evident that the performance of all the steel dish reflectors is comparable, but steel dish-2 has a slight edge over others in the 500 Hz-8 khz frequency range. Similarly, Fig. 7 shows a comparison of the plastic dish reflectors and it is evident that plastic dish-1 provides the best gain in the range of 2 khz- 8 khz. Interestingly plastic dish-3 due to its flat geometry is the most optimum in the range of 500 Hz-2 khz. Hence,

4 (a) microphone. If we compare the theoretical gain characteristics of Fig. 1 to Fig. 7(c), we can observe that experimental gain of the parabolic dishes does follow a similar pattern to the theoretical gain in the frequency range of 0-4 khz. The reason may be due to the design characteristics of the microphone which is primarily designed for close talking speech. All the frequency response measurements are made in the anechoic chamber. E. Directivity and Frequency Response in Reverb Condition (c) Fig. 6. Comparative analysis of different reflectors in terms of sensitivity and directivity (a) steel dish reflectors and no reflector setup plastic dish reflectors and no reflector setup (c) steel dish-2 and plastic dish-1 with no reflector setup (a) Fig. 8. (a) Sensitivity and directivity comparison in reverberant condition PSD of ambient noise in reverb condition as recorded by steel dish-2 reflector (a) (c) Fig. 7. Comparison of different reflectors in terms of frequency response (a) steel dish reflectors and no reflector setup plastic dish reflectors and no reflector setup (c) gain of steel dish 2 and plastic dish-1 over no reflector setup depending on the nature of sounds that we focus on, we can optimize the reflector. Finally, in Fig. 7(c), we have compared the gain of two best reflectors from Fig. 7(a) and. The gain has been estimated by calculating the improvement of the received signal power by a reflector dish over the no reflector To evaluate the performance of the reflector dishes in reverberant condition, we conducted the experiment in a conference room with dimensions of 8m 3m. We followed the same procedure as described in Section III. The resulting directivity pattern of steel dish-2 and plastic dish-1 in comparison to no reflector microphone has been shown in Fig. 8(a). dish-2 gives the best performance with a gain of 11 db with respect to a no reflector microphone, at 0 azimuth. We can also infer from Fig. 8(a) that the reflector dishes have improved the sensitivity between 30 to 330 azimuth angle, whereas the back-lobe rejection between 150 to 210 azimuth angle which are similar to no reflector microphone. Fig. 8(a) also shows that the use of reflector dish has changed the omni-directional characteristic of the microphone to somewhat directional. The Fig. 8 shows the power spectral density of received ambient noise using the steel dish-2. Here we can notice that the ambient noise is close to white in the frequency region of interest. The main source of noise is recording noise. IV. REFLECTOR MICROPHONE PERFORMANCE IN LARGE PREMISES Fig. 9(a) shows the approximate layout of our signal processing building, where we have installed the reflector dish

5 (a) Fig. 9. (a) Layout of premises monitoring setup sound source tracking using steel dish-2 and without the reflector for premises activity monitoring and moving sound source tracking. The microphone with the reflector dish (S2R) and no reflector microphone (NRM) are deployed besides each other. The sound source is chosen as white noise which moves from position SP to ML and then from position ML to EP as shown in the plan of Fig. 9(a). The real-time recording is done in a wireless environment with the help of Raspberry pi (NA and NB). The audio is streamed through wifi and stored in a central server (SR). The audio files as recorded by both the setups are shown in Fig. 9. The figure clearly shows a significant improvement of sound acquisition in one direction whereas a substantial rejection in the perpendicular direction. Therefore, the reflector dish setup has provided us with an opportunity to leap into the realm of spatial as well as temporal signal processing. From Fig. 9, we can see that in the case of no reflector microphone setup, a sound source moving from position SP to ML and ML to EP would generate the same signal level pattern in both halves. Hence, it would not be possible to ascertain whether the source is moving from SP to EP or vice-versa. In the case of reflector dish setup the pattern is not symmetric and therefore we can infer the movement of the sound source in the corridor in the direction of SP to EP. We can also infer the direction of movement by estimating short time energy (db) at location SP, ML and EP. Let these be denoted by E SP, E ML and E EP respectively. In the case of no reflector microphone, the difference of E ML to E SP is 12 db and E ML to E EP is 11.5 db which is somewhat similar. However, in the case of reflector microphone set up the E ML to E SP difference is equal to 14 db and the E ML to E EP difference is 18 db. This result is attributable to the directive gain of the reflector microphone at 0 which clearly is 4 db more than that is attained at 90. The dip in the short time energy is a clear indication that the source was traveling in the direction SP to ML and then to EP. This deduction about the temporal-spatial information about the source would not be possible in the prior case where the short time energy ratios are equal. Therefore we can justify the preference of reflector microphone setup over the omni-directional microphone in Indoor Audio Monitoring. V. CONCLUSION We have shown an economic way of converting an omniclose talking microphone to a directional longer range microphone using an acoustic reflector. The directional sensitivity pattern changes with the dish shape and even non-parabolic reflector would be useful for certain applications. Our results show that a reflector microphone setup gives 60 of directionality and hence we can combine a number of such reflector microphone setups to gather nationalized spatial and temporal information. The reflector setup can also be used as an inexpensive alternative to line array microphone. ACKNOWLEDGEMENT We thank Robert Bosch Center for Cyber Physical Systems (RBCCPS), IISc, for providing us with the essential funds. We would like to thank Mr. Srikanth C., Ms. Priyadarshini Kannan and Mr. Sangeeth Jayan for their valuable contribution in performing experiments in reverb/anechoic room. REFERENCES [1] J. W. Rayleigh, The theory of sound, Macmillan, vol. 27, [2] L. Cremer, Die wissenschaftlichen grundlagen der raumakustik, Hirzel Stuttgart, [3] Y. Nishida, T. Hori, T. Suehiro and S. Hirai, Monitoring of breath sound under daily environment by ceiling dome microphone, Systems, Man, and Cybernetics, 2000 IEEE International Conference on, vol. 3, pp , [4] H. Aumann, T. Russell, and N. W. Emanetoglu Comparison of a Small Parabolic Reflector for Use with an Acoustic and a Radar Microphone, IEEE International Symposium on Antennas and Propagation & USNC/URSI National Radio Science Meeting, vol. 3, pp , [5] J. W. Benson, Y. Liang and G. W. Swenson A baffle-type directional microphone, The Journal of the Acoustical Society of America, vol. 95, no. 5 pp , [6] X. Dai, J. Zhu, Yi-Te Tsai, Use of parabolic reflector to amplify inair signals generated during impact-echo testing, The Journal of the Acoustical Society of America, vol. 130, no. 4 pp , [7] Z. Chen, R. C. Maher Parabolic Dish Microphone System, Montana State University, 2012.