Proceedings of Meetings on Acoustics

Transcription

1 Proceedings of Meetings on Acoustics Volume 19, ICA 213 Montreal Montreal, Canada 2-7 June 213 Psychological and Physiological Acoustics Session 1pPPa: Binaural Hearing and Binaural Techniques I 1pPPa11. Measuring pressure and particle velocity along the human ear canal Marko Hiipakka* *Corresponding author's address: Signal Processing and Acoustics, Aalto University, Espoo, 215, Southern Finland, Finland, Marko.Hiipakka@aalto.fi A non-invasive method of measuring or estimating accurately the head-related transfer functions (HRTFs) and headphone transfer functions (HpTFs), i.e., the pressure at the eardrum rather than at the blocked ear canal entrance is called for. In this work, a miniature-sized acoustic pressure-velocity sensor is used to measure both pressure and velocity along the ear canals of human test subjects. The measurements are used to study the applicability of a recently proposed method of estimating the pressure at the eardrum from pressure-velocity measurements made at the ear canal entrance. The measurement results are compared to results from computational modeling with human ear canal parameters. In addition, the effect of the PU-sensor itself on the pressure at the eardrum is studied. It is shown that the estimation method is reliable and accurate for most human subjects. The diameter and the shape of the ear canal affect the results in such a way that the best results are obtained with wide and straight ear canals. It is concluded that the estimation method facilitates the obtaining of individual HRTFs and HpTFs at the eardrum using non-invasive measurements. Published by the Acoustical Society of America through the American Institute of Physics 213 Acoustical Society of America [DOI: 1.11/ ] Received 23 Jan 213; published 2 Jun 213 Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 1

2 INTRODUCTION The method of estimating the pressure at the eardrum ( ) discussed in this paper is based on measurements of acoustic pressure and particle velocity at the ear canal entrance and along the ear canal [1][2][3]. Other methods to achieve the same goal have been presented, but none of these have utilized direct measurements of particle velocity. In most of the solutions presented, however, estimates of the volume velocity or impedance at the ear canal entrance have been used to solve. One important motivation of an accurate estimation method is in the unreliability and risks related to probe microphone measurements at the eardrum. It is common to first solve the individual ear canal parameters, i.e., the area function [4, 5] and the length [] of the ear canal and the impedance of the eardrum in order to estimate with physics-based computational modeling [7]. Hudde has presented a variety of methods to solve the ear canal parameters, such as the estimation of the area function by sound pressure measurements [8], and measurement of the eardrum impedance of human ears [9] using, for instance, an acoustic measuring head [1]. In addition, sound intensity and the forward wave component of the sound pressure wave have been used to estimate [11]. These methods include separating the incident acoustic intensity in the ear canal from the reflected intensity [], or using forward pressure level to avoid the effect of standing waves [13]. The goal of the research presented in [1] and [3] was to find a solution for measuring HRTFs with the eardrum as the point of reference without the need to resort to probe or probe tube microphones. It was suggested that the task could be accomplished through measurements with the Microflown PU sensor, which is a relatively new device able to measure acoustical particle velocity by utilizing hot-wire anemometer technology [14]. In this paper, the goal is to further validate the applicability of the energy-based estimation method presented earlier. The first question discussed is related to the positioning of the PU sensor in the human ear canal. According to the principle presented in [3], the PU sensor can be placed at any depth in the ear canal without losing estimation accuracy. The theory is tested by running several estimations using the pressure and the velocity obtained from various depths in the ear canal. The second subject of interest is the magnitude of the particle velocity close to the eardrum, since the functionality of the estimation method in question requires the velocity to be almost insignificant in comparison to pressure at close proximity to the eardrum. Thirdly, the effect of the PU sensor at the ear canal entrance on is studied. The new results presented here are preliminary by nature. PREVIOUS STUDIES ON THE TOPIC The idea of using a PU sensor for estimating the pressure at the eardrum was first tested using a standard Microflown PU sensor and an ear canal simulator with constant cross-sectional area [3]. After careful analysis of the measured pressure and velocity signals, the conclusion was that computation of the energy density at the entrance of the simulator could be used to estimate the magnitude of the pressure at the eardrum of the simulator. The idea is based on the assumption that very close to the eardrum, most of the sound energy is concentrated on the pressure component due to the high impedance of the eardrum. Since the structure of the standard version of the PU probe is such that it cannot be used for measurements inside a human ear canal, two custom made probes were designed in cooperation with Microflown technologies (Fig. 1). The probes were equipped with a mesh that protects the very thin platinum wires of the hot-wire anemometer from breakage by contact with hair. Measurements with 25 human subjects were carried out in order to test whether the method was applicable to human ear canals, the length and cross-sectional area of which were unknown. The estimated pressures at the eardrums of human subjects were compared to actual microphone Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 2

3 FIGURE 1: On the left: a standard Microflown PU probe and a custom-made ear canal simulator. On the right: a custom-made Microflown PU probe. [2] measurements at the eardrums. In a study by Takanen et al. [15] the method was verified through a listening test where the audibility of coloration caused by different HRTF filter design methods were investigated. The comparisons between the eardrum measurements and the estimates showed that the magnitude of the pressure at the eardrum of an ear canal simulator, a dummy head, and human test subjects can be estimated using a pressure-velocity measurement at the entrance of the ear canal. The method also works with various eardrum impedances and non-uniform cross-sectional ear canal areas and not only with completely rigid eardrums and straight ear canals. The eardrum pressure can be estimated using the transmission line equations, too, but the energybased estimation is applicable without accurate information on the length and cross sectional area of the ear canal. The magnitudes of the HRTF filters designed using the PU measurements proved to be closer to those designed from eardrum probe measurements than the ones designed using blocked ear canal HRTF measurements. The listening test confirmed the findings as the filters designed using the energy-based estimation caused less coloration than the method utilizing blocked ear canal measurements. Part of these findings can be explained with the fact that the open-back circumaural headphones used in the study do not have completely free-air equivalent coupling to the ear. Regarding HRTF filter design, the approach presented proved to be at least as effective as the traditional blocked ear canal method. METHODS Computational models of individual ear canals were constructed for the study. The various parameters needed for the modeling were estimated separately for each individual ear canal. The eardrum impedance model (Z D ) used in the study is based on unpublished pressure and velocity measurements in human ear canals as well as at the eardrum of an ear replica with an artificial eardrum presented in [1]. The ear canal was as a transmission line with variable cross-sectional diameter. The area function of the ear canal was determined with the help of castings of human ear canals as well as images of human ear canals obtained by magnetic resonance imaging. One example of the radius function of an individual ear canal is shown in Fig. 2. The transmission line was divided to sections with the length of 1 mm each and the pressure and velocity at each point could be computed separately. The impedance seen towards the ear canal at the ear canal entrance (Z ECE ) was computed from the pressure and velocity at the entrance when a pressure source at the eardrum was feeding the ear canal. For a simple tube-like ear canal simulator in free field [1], the acoustic impedance (Z S ) of an external pressure source is equivalent to the radiation impedance of the tube opening due to the reciprocity principle. However, to obtain an source impedance model (Z S ) for human ear canals, the radiation impedance was adjusted to better fit impedances seen in previous measurements. The Thévenin source model pressure (P S ) was obtained from individual blocked-ear-canal HRTF measurements (azimuth, elevation ) made earlier for [1]. Having the blocked HRTF measurement as (P S ) adds the effect of torso, the head, and the pinna to the model, but the ear canal still needs to be separately and connected to the pressure source. Hence, the Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 3

4 Radius [mm] Length [mm] FIGURE 2: The radius function of an individual ear canal from the entrance to the beginning of the eardrum. When the PU probe is placed at the ear canal entrance, it effectively diminishes the cross-sectional diameter of the canal. radiation of pressure (P S ) into the open ear canal is computed as Z ECE P E = P S. (1) Z ECE + Z S A detailed presentation of the energy-based estimation method discussed here can be found in [1]. The method is based on the assumption that the total acoustic energy density, which is the sum of the kinetic energy density and potential energy density, is preserved in an acoustic waveguide with constant cross sectional area. D = D k + D p = 1 2 ρ u 2 + p 2 2ρc 2 (2) Because of the relatively high impedance of the eardrum compared, e.g., to wave impedance and to Z S, the particle velocity is very small at close proximity to the eardrum, hence, 1 2 ρ u 2 << p 2 2ρc 2, (3) which means that most of the energy density is concentrated on the potential energy component. If pressure and particle velocity at the ear canal entrance or any other point in the ear canal is known, the magnitude of can be estimated using Eq. () or (7) in [1]. Since the estimation method yields only the magnitude of the pressure, the phases of the velocity and the pressure used in the estimate can be disregarded. When the particle velocity is measured at the open ear canal entrance, information of the cross sectional area and the length of the ear canal are not used as parameters in the energy-based estimation method. RESULTS First the measurement and estimation results for one test subject are shown in Fig. 3 to facilitate comparison with the corresponding responses. With good estimates of the individual ear canal parameters the model yields responses very similar to those obtained from actual anechoic measurements. In this case the length of the ear canal is 3 mm. The shape of the simulated ear canals proved to affect the results of the energy-based estimation method as those ear canals with largely varying cross sectional areas yielded less accurate results for the estimation method. At the ear canal entrance the particle velocity (scaled by ρc) is relatively large and carries plenty of information about the pressure at the eardrum as depicted in Fig. 3 and Fig. 4. In both figures, the position of the (real and simulated) particle velocity measurement is 1 mm inside the ear canal. The measurement point of pressure (at 8 mm), however, is 2 mm outwards from the point of the velocity measurement due to the structure of the PU sensor used (See. Fig.1). Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 4

5 measured estimated U E measured P E measured 5 1k Frequency [Hz] 1k 15k FIGURE 3: Measured and estimated pressure at the eardrum of one subject. The estimation is made using measured pressure P E and velocity U E from the open ear canal entrance. Here U E is scaled by ρc. Direction of sound incidence is azimuth, elevation. estim. U E P E 5 1k Frequency [Hz] 1k 15k FIGURE 4: Modeled at the eardrum of one subject; estimated obtained using pressure P E and velocity U E from the open ear canal entrance. U E is scaled by ρc. In Fig. 5, both the pressure and the velocity are simulated at a point halfway into the ear canal (15 mm). The dip in the pressure frequency response, which is caused mainly by the sound reflection from the eardrum, has shifted from 4 khz to khz. At a distance of 1 mm from the eardrum, in Fig., the velocity has started to drop significantly at frequencies up to - 7 khz. Finally, in Fig. 7, the velocity at a distance of 1 mm from the eardrum has dropped at frequencies up to approximately 1 khz. This behavior of the pressure and velocity components along the ear canal bolsters the applicability of the estimation method discussed in this paper. The estimates, which are obtained using pressure and particle velocity at various points in the ear canal are similar in all the results presented here. Hence, it is shown that the insertion depth of the PU probe does not affect the estimation result. The PU sensor in the ear canal effectively diminishes the cross sectional area of the ear canal at its position (See Fig. 2). The results depicted in Fig. 8 show that this has a significant effect on the pressure at the eardrum at frequencies between 3 and 7 khz. With a very narrow ear canal the effect of the PU sensor becomes more significant as the sensor blocks a relatively larger portion of the cross sectional area at the ear canal entrance. Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 5

6 estim. U 2 P 2 5 1k Frequency [Hz] 1k 15k FIGURE 5: Modeled at the eardrum of one subject; estimated obtained using pressure P 2 and velocity U 2 from a point halfway into the ear canal. U 2 is scaled by ρc. estim. U 3 P 3 5 1k Frequency [Hz] 1k 15k FIGURE : Modeled at the eardrum of one subject; estimated obtained using pressure P 3 and velocity U 3 from a point 1 mm from the eardrum. U 3 is scaled by ρc. CONCLUSIONS The new method of measuring the pressure at the human eardrum presented earlier in [1] and [2] is a safe and reliable alternative to probe microphone measurements of pressure at the eardrum. In this study, measurements made with a miniature-sized acoustic pressure-velocity sensor along the ear canals of human test subjects were used to study the applicability of the method of estimating the pressure at the eardrum. Computational models of human ear canals were used to study the behavior of the pressure and velocity components along the ear canal, as well as the effect of the PU sensor on the pressure at the eardrum. It was concluded that the choice of measurement point along the ear canal does not significantly affect the result of the estimation method. However, blocking of the ear canal with the PU sensor should be minimized. The new results presented here strongly support that the energy-based estimation method is a reliable non-invasive method of obtaining individually the pressure at the eardrum for, e.g., binaural reproduction or audiological purposes. Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page

7 estim. U D 2 5 1k Frequency [Hz] 1k 15k FIGURE 7: Modeled at the eardrum of one subject; estimated obtained using pressure 2 and velocity U D from a point 1 mm from the eardrum. U D is scaled by ρc Without probe With probe 5 1k Frequency [Hz] 1k 15k FIGURE 8: Modeled at the eardrum of a subject in two different situations: with and without the PU probe at the ear canal entrance. Ear canal length, in this case, is 2 mm. Ear canal radius function is depicted in Fig. 2. ACKNOWLEDGMENTS The research leading to these results has received funding from the European Research Council under the European Community s Seventh Framework Programme (FP7/27-213) / ERC grant agreement n o [24453]. REFERENCES [1] M. Hiipakka, T. Kinnari, and V. Pulkki, Estimating head-related transfer functions of human subjects from pressure-velocity measurements, Journal of the Acoustical Society of America 131, (2). [2] M. Hiipakka, Estimating pressure at the eardrum for binaural reproduction, Ph.D. thesis, Aalto University, Dept. of Signal Processing and Acoustics (2). [3] M. Hiipakka, M. Karjalainen, and V. Pulkki, Estimating pressure at eardrum with pressure-velocity measurement from ear canal entrance, in IEEE Workshop on Applications of Signal Processing to Audio and Acoustics, (New Paltz, NY, USA) (29). [4] M. Stinson and B. Lawton, Specification of the geometry of the human ear canal for the prediction of sound-pressure level distribution, Journal of the Acoustical Society of America 85 (1989). Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 7

8 [5] D. Rasetshwane and S. Neely, Inverse solution of ear-canal area function from reflectance, Journal of the Acoustical Society of America 13, (211). [] J. Chan and C. Geisler, Estimation of eardrum acoustic pressure and of ear canal length from remote points in the canal, Journal of the Acoustical Society of America 87, (199). [7] H. Hudde, A. Engel, and A. Lodwig, Methods for estimating the sound pressure at the eardrum, Journal of the Acoustical Society of America 1, (1999). [8] H. Hudde, Estimation of the area function of human ear canals by sound pressure measurements, Journal of the Acoustical Society of America 73, (1983). [9] H. Hudde, Measurement of the eardrum impedance of human ears, Journal of the Acoustical Society of America 73, (1983). [1] H. Hudde, A. Lodwig, and A. Engel, A wide-band precision acoustic measuring head, Acustica 82, (199). [11] S. Neely and M. Gorga, Comparison between intensity and pressure as measures of sound level in the ear canal, Journal of the Acoustical Society of America 14, (1998). [] B. Farmer-Fedor and R. Rabbitt, Acoustic intensity, impedance and reflection coefficient in the human ear canal, Journal of the Acoustical Society of America 1, 2 (22). [13] R. Scheperle, S. Neely, J. Kopun, and M. Gorga, Influence of in situ, sound-level calibration on distortion-product otoacoustic emission variability, Journal of the Acoustical Society of America 4, (28). [14] H.-E. de Bree, An overview of microflown technologies, Acta Acustica united with Acustica 89, (23). [15] M. Takanen, M. Hiipakka, and V. Pulkki, Audibility of coloration artifacts in HRTF filter designs, in Proc. 45th AES International Conference (Espoo, Finland) (2). [1] M. Hiipakka, M. Tikander, and M. Karjalainen, Modeling of external ear acoustics for insert headphone usage, Journal of the Audio Engineering Society 58, (21). Proceedings of Meetings on Acoustics, Vol. 19, 519 (213) Page 8