Development of A Steerable Stereophonic Parametric Loudspeaker

Transcription

1 Development of A Steerable Stereophoni Parametri Loudspeaker Chuang Shi, Hideyuki Nomura, Tomoo Kamakura, and Woon-Seng Gan Department of Eletrial and Eletroni Engineering, Kansai University, Osaka, Japan r148005@kansai-uajp Department of Eletroni Engineering, The University of Eletro-Communiations, Tokyo, Japan hnomura@ueajp Center for Industrial and Governmental Relations, The University of Eletro-Communiations, Tokyo, Japan kamakura@eeueajp Shool of Eletrial and Eletroni Engineering, Nanyang Tehnologial University, Singapore ewsgan@ntuedusg Abstrat The parametri loudspeaker is a type of diretional loudspeakers making use of the nonlinear aousti effets The past studies to reprodue the three-dimensional audio ontents with a pair of the parametri loudspeakers have demonstrated satisfatory performane In this paper, the steerable parametri loudspeakers are proposed to reloate the sweet spot to follow the head movement of the listener Although the spatial aliasing effets are observed in the steerable parametri loudspeaker, they an be onverted to generate multiple sound beams simultaneously A new ase of the grating lobe elimination, namely the over elimination, is studied to extend the ontrollable level differene between the two sound beams The simulation results to ompare the equal and Chebyshev weights are also presented in this paper I INTRODUCTION The parametri loudspeaker is a type of diretional loudspeakers that transmits a narrow sound beam with a relatively smaller emitter size as ompared to a onventional loudspeaker or loudspeaker array [1] As an appliation of the parametri transmitting array in air, the parametri loudspeaker generates a virtual endfire array from the parametri array effet Hene, the parametri loudspeaker is understood to be a ombination of the eletroni hardware and the nonlinear aousti effets When the ultrasoni emitter of the parametri loudspeaker is bloked, it is muted, beause there is no virtual soure formed in the air [2] A ommon struture of the parametri loudspeaker is shown in Fig 1 [3] In this struture, a bandpass filter is introdued to inrease the reprodued sound pressure level and to redue the intermodulation distortions, whih is similar to the bandpass filter used in the telephone systems The driving iruit onsists of an amplifier and a proessor or a hardware modulator Preproessing and modulation methods are often arried out in the driving iruit The filtered audio signal will be modulated on an ultrasoni arrier There is no demodulator required at the loation of the listener, beause of the self-demodulation effet [4] The ultrasoni emitter has the resonane frequeny equaling to the arrier frequeny used in the modulator The arrier frequeny and emitter size determine the absorption and Rayleigh distanes The absorption distane gives the length Fig 1 Common struture of the parametri loudspeaker [3] of the virtual endfire array, while the Rayleigh distane defines the near field boundary of the ultrasoni emitter In the early studies of the parametri loudspeaker, the Rayleigh distane was designed to be longer than the absorption distane, and the observation points were plaed beyond the Rayleigh distane [5] In the steerable parametri loudspeaker, the ultrasoni emitter onsists of hundreds of the piezoeletri erami transduers (PZTs) They are grouped into several hannels and driven by individual amplifiers [6] The phased array beamsteering is adopted in the steerable parametri loudspeaker to generate a ontrollable sound beam [7] A diagram of the beamsteering struture is shown in Fig 2 Individual delays and weights an be applied to the arrier and sideband frequenies The spatial aliasing effets are adverse to the steerable parametri loudspeaker [8] Aording to the Nyquist riterion, the ultrasoni emitter onsisting of the PZTs resonating at 40 khz requires the interhannel spaing being less than 85 mm to prevent the ourrene of grating lobes This requirement is diffiult to be arried out, for most of the PZTs have the diameter of 10mm or 16 mm Some experimental onfigurations of the ultrasoni emitter are shown in Fig 3 There are three potential approahes to deal with the spatial aliasing effets in the steerable parametri loudspeaker Firstly, the ompat onfiguration an be adopted in the ultrasoni APSIPA APSIPA 2014

2 emitter as shown in Fig 3(d) [9] In the ompat onfiguration, the interhannel spaing is redued to only half of the diameter of the PZTs An implementation of the ompat onfiguration has been arried out in the steerable parametri loudspeaker for ative noise ontrol [10] In ontrast to the olumn onfiguration using the same number of PZTs in every hannel as shown in Figs 3(a)-3(), the ompat onfiguration ontains different numbers of the PZTs in the odd and even hannels Hene, a larger number of the PZTs is usually required in the ompat onfiguration to ensure the output power is equally spread aross all the hannels [11] Seondly, the miroeletromehanial (MEMS) tehnology helps to redue the size of the PZTs A MEMS PZT array has been fabriated and examined [12] The diameter of every ultrasoni transduer in this array is as small as one eighth the diameter of the onventional PZTs The piezoeletri thin film is speially designed to improve the aousti effiieny of the MEMS PZT [13] Therefore, the reprodued sound pressure level is not muh ompromised by the redued emitter size Thirdly, the grating lobe elimination has been proposed in theory and validated by experiments [14] When the interhannel spaing is too wide to fulfill the Nyquist riterion, grating lobes are observed at the arrier and sideband frequenies However, there is little or no spatial aliasing observed at the differene frequeny The steerable parametri loudspeaker an be designed to take advantage of the grating lobe elimination, so the olumn onfiguration in Fig 3(a) an be free from the spatial aliasing effets In this paper, the spatial aliasing effets are made use of to generate two sound beams Thus, a stereophoni reprodution method is proposed to transmit different audio ontents to two diretions simultaneously from only one steerable parametri loudspeaker The proposed method will benefit the nonwearable stereophoni sound reprodution [15] [17] For example, the steerable parametri loudspeaker an be installed in the portable devie, where two sound beams an be steered to the left and right ears of the listener and the stereophoni audio ontents an be reprodued without earphones Fig 2 Beamsteering struture of the steerable parametri loudspeaker, where SSBAM-SC is the abbreviation for the single sideband amplitude modulation suppressed arrier II T HEORY A uniform linear array onsisting of M hannels has the equal interhannel spaing d The input signal is assumed to represent a plane wave at the angular frequeny of ω and wavenumber of k When the weight wm and delay τm are applied to the mth hannel for m = 0, 1,, M 1, the output of eah hannel observed in the far field at the inidene angle of θ is written as Cm (θ) = wm exp (jmdk sin θ + jωτm ), where j = 1 is the imaginary unit The delay amount an be designed as τm (θ0 ) = md sin θ0, (1) (2) Fig 3 Photos of ultrasoni emitters using different onfigurations, where the interhannel spaing is denoted as d where θ0 is the steering angle, and is the speed of sound Thereby, the beampattern of this linear array is alulated by M 1 X 2πf D (θ, f, θ0 ) = wm exp j md (sin θ sin θ0 ) m=0 (3) As shown in Fig 2, the delays τm1 and τm2 are applied to the first primary frequeny f1 and the seond primary frequeny f2 = f1 +fd, respetively The two groups of delays

3 Fig 4 Four ases of grating lobe elimination in the olumn onfiguration (d = 125m), where the primary waves at 40 khz and 44 khz are plotted in blak and red, respetively; and the differene frequeny at 4 khz is plotted in blue Fig 5 Four ases of grating lobe elimination in the olumn onfiguration (d = 15m), where the primary waves at 40 khz and 44 khz are plotted in blak and red, respetively; and the differene frequeny at 4 khz is plotted in blue result in two steering angles of θ1 and θ2, respetively Based on the produt diretivity priniple [18] [20], the diretivity of the differene frequeny fd an be omputed in a simplified expression as so that the mainlobe and grating lobe our symmetrially to 0 as well This beamsteering struture is alled the symmetri struture [16] Based on the produt diretivity priniple, the differene frequeny has its mainlobe and grating lobe being symmetri to 0 The dual beam generation an transmit the same audio ontents to two different diretions simultaneously From Figs 4(a) to 5(a), the angular separation between the mainlobe and grating lobe of the differene frequeny is narrowed from 383 to 318, beause the inreased interhannel spaing leads to the redued spatial aliasing periods of the primary frequenies The seond ase, namely, the stereophoni beam generation, is shown in Figs 4(b) and 5(b) In the stereophoni beam generation, the steering angle of the seond primary frequeny is adjustable to ontrol the level differene between the two sound beams [15] In omparison with the dual beam generation, the mainlobe of the seond primary frequeny is steered loser to the mainlobe of the first primary frequeny Here, an example of the steering angle of the seond primary frequeny is given by 1 θ2 = sin 1 + sin 1 (7) 2 2f2 d Dd (θ) = D (θ, f1, θ1 ) D (θ, f2, θ2 ) (4) In the diretivity of the differene frequeny, the angular diretions of the mainlobe and the first grating lobe are denoted as θa and θb, respetively Furthermore, an envelope method an improve the auray of the produt diretivity priniple [20], whih applies the spline interpolation between the loal maxima of (4) III F OUR C ASES OF G RATING L OBE E LIMINATION In this setion, the olumn onfigurations in Figs 3(b) and 3() are adopted, where M = 8 The interhannel spaings of them are given by 125 m and 15 m, respetively Both the interhannel spaings are larger than the wavelength of the 40 khz wave The speed of sound is estimated at 344 m/s The primary frequenies are seleted at 40 khz and 44 khz, and the differene frequeny at 4 khz is expeted to be generated The steering angle of the first primary frequeny is given by θ1 = sin 1 (5) The mainlobe and grating lobe are steered to be symmetri to 0 [16] The equal weights are used When the seond primary frequeny is steered to different angles θ2, four ases of grating lobe elimination an be derived from (4) Firstly, the ase of dual beam generation is shown in Figs 4(a) and 5(a) In the dual beam generation, the steering angle of the seond primary wave is given by θ2 = sin 1, (6) 2f2 d The third ase is the steerable parametri loudspeaker, where the steering angles of the two primary frequenies equal to the intended steering angle of the differene frequeny, ie θ2 = θ1 = sin 1 (8) In Figs 4() and 5(), the mainlobes of the primary frequenies oinide with eah other However, the grating lobes of the primary frequenies our apart, due to the differene in the spatial aliasing periods of the primary frequenies The resultant grating lobe of the differene frequeny will be eliminated

4 The high differene frequeny leads to the full elimination of grating lobes, while the low differene frequeny results in the partial grating lobe elimination only [14] In this paper, the fourth ase of grating lobe elimination is derived, whih is alled the over elimination as demonstrated in Figs 4(d) and 5(d) The maximum level differene between the two sound beams in the stereophoni beam generation is used to be onstrained by the partial grating lobe elimination [15] The over elimination provides a signifiant inrement of the maximum level differene between the two sound beams generated from one steerable parametri loudspeaker IV SIMULATION RESULTS The intersetion funtion was originally proposed in [8] and experimentally validated in [14] With the variables defined in this paper, the original intersetion funtion is rewritten as I old (f d ) = max [D (θ, f 1, 0) D (θ, f 2, θ )], (9) θ <θ<3θ where θ = sin 1 f dd (10) In order to omparatively study the four ases of grating lobe elimination, the intersetion funtion has to be modified as max [D (θ, f 1, 0) D (θ, f 2, θ )] θ I new (f d, φ) = <θ<3θ max [D (θ, f, 1, 0) D (θ, f 2, φ)] φ<θ<3φ (11) where φ = θ 1 θ 2 is the angular separation between the mainlobes of the primary frequenies, and ( θ = sin 1 sin φ + f 2 d ) (12) f 1 d is the angular separation between the grating lobes of the primary frequenies The modified intersetion funtion an alulate the maximum level differene between the two sound beams generated from one steerable parametri loudspeaker The results obtained from (9) and (11) using four sets of weights are plotted in Fig 6 The four sets of weights inlude one set of equal weights and three sets of Chebyshev weights resulting in sidelobe attenuations of 13 db, 20 db, and 30 db It is observed in Fig 6 that the proposed over elimination has resulted in a wider range of level differene between the two sound beams when ompared with the previous stereophoni beam generation, in spite of the set of weights adopted In the previous method, the set of Chebyshev weights with the largest sidelobe attenuation results in the worst performane in terms of the maximum level differene However, the same set of Chebyshev weights demonstrates the best performane when using the proposed over elimination Although the over elimination an improve the maximum level differene, the mainlobe level is ompromised for the low differene frequeny Similarly, the dual beam generation ahieves the same levels of the two sound beams, but the mainlobe level is ompromised for the high differene frequeny In Fig 7, the mainlobe levels are ompared between the over elimination and the dual beam generation to demonstrate Fig 6 Maximum level differene between the two sound beams generated in the over elimination and the stereophoni beam generation, where blue lines represent the over elimination; and red lines represent the stereophoni beam generation Fig 7 Mainlobe level in the over elimination and the dual beam generation, where blue lines represent the over elimination; and red lines represent the dual beam generation elimination the power effiieny of the steerable stereophoni parametri loudspeaker The input power is kept onsistent The set of Chebyshev weights with the largest sidelobe attenuation results in the lowest power effiieny to generate the low differene frequeny, but the highest power effiieny to generate the high differene frequeny It shows that a wider range of the level differene between the two sound beams would request higher power input of the steerable stereophoni parametri loudspeaker In the future, we will arry out aousti measurements to find the tradeoff settings and examine the sound quality by subjetive assessments

5 V CONCLUSIONS Complementing to the three known ases of grating lobe elimination, the fourth ase, namely the over elimination, has been demonstrated in this paper The intersetion funtion, whih used to desribe the level of grating lobe elimination, has been modified to evaluate the maximum level differene between the two sound beams generated from one steerable parametri loudspeaker By aounting for the over elimination, the steerable stereophoni parametri loudspeaker has ahieved a wider range of the level differene between the two sound beams as ompared to the previous method Four sets of weights have also been ompared in the simulation The set of Chebyshev weights with the largest sidelobe attenuation an lead to the best performane of the stereophoni beamsteering However, there is a remaining issue to improve the power effiieny of the steerable stereophoni parametri loudspeaker [15] C Shi, H Nomura, T Kamakura, and W S Gan, Spatial aliasing effets in a steerable parametri loudspeaker for stereophoni sound reprodution, IEICE Trans Fund Eletron Commun Computer Si, vol E97-A, no9, pp , 2014 [16] C Shi, E L Tan, and W S Gan, Hybrid immersive three-dimensional sound reprodution system with steerable parametri loudspeakers, in Pro 21st Int Congr Aoust, Montreal, Canada, 2013 [17] S Aoki, M Toba, and N Tsujita, Sound loalization of stereo reprodution with parametri loudspeakers, Applied Aoust, vol 73, no 12, pp , 2012 [18] C Shi and W S Gan, Produt diretivity models for parametri loudspeakers, J Aoust So Am, vol 131, no 3, pp , 2012 [19] C Shi and W S Gan, Modeling the diretivity of parametri loudspeaker, in Pro 19th Int Symp Nonlinear Aoust, Tokyo, Japan, 2012 [20] C Shi, H Nomura, T Kamakura, and W S Gan, An envelope method for improving the produt diretivity models of the parametri loudspeaker, IEICE Teh Report, vol 113, no 412, pp 7-10, 2014 ACKNOWLEDGMENT This work is partially supported by MEXT-Supported Program for the Strategi Researh Foundation at Private University, REFERENCES [1] M Yoneyama, J Fujimoto, Y Kawamo, and S Sasabe, The audio spotlight: An appliation of nonlinear interation of sound waves to a new type of loudspeaker design, J Aoust So Am, vol 73, no 5, pp , 1983 [2] C Shi and Y Kajikawa, An introdution to parametri array loudspeaker and its appliation in ative noise ontrol, in Pro 1st Joint Symp Signal Proess Control Syst, Jungli, Taiwan, June 2014 [3] C Shi, Investigation of the steerable parametri loudspeaker based on phased array tehniques, PhD Thesis, Nanyang Tehnologial University, Singapore, 2013 [4] H O Berktay, Possible exploitation of nonlinear aoustis in underwater transmitting appliations, J Sound Vib, vol 2, no 4, pp , 1965 [5] P J Westervelt, Parametri aousti array, J Aoust So Am, vol 35, no 4, pp , 1963 [6] S Takeoka and Y Yamasaki, Aousti projetor using diretivity ontrollable parametri loudspeaker array, in Pro 20th Int Congr Aoust, Sydney, Australia, 2010 [7] D Olszewski, F Prasetyo, and K Linhard, Steerable highly diretional audio beam loudspeaker, in Pro Interspeeh 2005, Lisbon, Portugal, 2005 [8] C Shi and W S Gan, On grating lobe elimination of differene frequeny in parametri loudspeaker, in Pro2010 APSIPA Annu Summit Conf, Singapore, 2010 [9] W S Gan, J Yang, K S Tan, and M H Er, A digital beamsteerer for differene frequeny in a parametri array, IEEE Trans Audio Speeh Lang Proess, vol 14, no 3, pp , 2006 [10] N Tanaka and M Tanaka, Ative noise ontrol using a steerable parametri array loudspeaker, J Aoust So Am, vol 127, no 6, pp , 2010 [11] S Wu, M Wu, C Huang, and J Yang, FPGA-based implementation of steerable parametri loudspeaker using frational delay filter, Applied Aoust, vol 73, no 12, pp , 2012 [12] H Lee, D Kang, and W Moon, A miro-mahined soure transduer for a parametri array in air, J Aoust So Am, vol 125, no 4, pp , 2009 [13] Y Je, H Lee, and W Moon, The impat of miromahined ultrasoni radiators on the effiieny of transduers in air, Ultrasonis, vol 53, no 6, pp , 2013 [14] C Shi and W S Gan, Grating lobe elimination in steerable parametri loudspeaker, IEEE Trans Ultrason Ferroeletr Freq Control, vol 58, no 2, pp , 2011