Academia.eduAcademia.edu

Figure 23 – uploaded by Deepak Gala

See full PDFdownloadDownload figure
Table 8 Experimental results using the robotic platform: Orientation localization using 3D model (RMSE: difference between azimuth estimations using the 2D and 3D models, respectively)

Table 8 Experimental results using the robotic platform: Orientation localization using 3D model (RMSE: difference between azimuth estimations using the 2D and 3D models, respectively)

Related Figures (32)

Fig. 1 Interaural Time Delay (ITD) estimation between signals yj (t) and y2(t) using the cross-correlation technique The time difference of yj(t) and y2(f), ie., the ITD, is given by TA S arg max, Ry,,y,. The distance difference of the sound signal traveling to the two microphones is given
Fig. 1 Interaural Time Delay (ITD) estimation between signals yj (t) and y2(t) using the cross-correlation technique The time difference of yj(t) and y2(f), ie., the ITD, is given by TA S arg max, Ry,,y,. The distance difference of the sound signal traveling to the two microphones is given
Fig.2 Top view of the robot illustrating different angle definitions due to the rotation of the microphone array 3.1 Definitions As shown in Figs. 2 and 3, the acoustic signal generated by the sound source S' is collected by the left and right microphones, L and R, respectively. Let O be the center of the robot as well as the two microphones. The location of S is represented by (D,6,q@), where D is the distance between the source and the center of the robot, i.e., the 4 length of segment OS, 6 € [0, 5 | is the elevation angle defined as the angle between O'S and the horizontal plane, and g@ € (—z, 7] is the azimuth angle defined as the angle measured clockwise from the robot heading vector, p, to OS. Letting unit vector q be the orientation (heading) of the microphone array, 6 be the angle between p and q, and y be the angle between q and OS, both following a right hand rotation rule, we have
Fig.2 Top view of the robot illustrating different angle definitions due to the rotation of the microphone array 3.1 Definitions As shown in Figs. 2 and 3, the acoustic signal generated by the sound source S' is collected by the left and right microphones, L and R, respectively. Let O be the center of the robot as well as the two microphones. The location of S is represented by (D,6,q@), where D is the distance between the source and the center of the robot, i.e., the 4 length of segment OS, 6 € [0, 5 | is the elevation angle defined as the angle between O'S and the horizontal plane, and g@ € (—z, 7] is the azimuth angle defined as the angle measured clockwise from the robot heading vector, p, to OS. Letting unit vector q be the orientation (heading) of the microphone array, 6 be the angle between p and q, and y be the angle between q and OS, both following a right hand rotation rule, we have
Fig.3 3D view of the system for orientation localization
Fig.3 3D view of the system for orientation localization
Fig.4 The shaded triangle in Fig. 3. cosa = cos@ sin yw. Based on the far-field assumption in Section 2.2, we have where b is the distance between the two microphones, i.e. the length of the segment LR.
Fig.4 The shaded triangle in Fig. 3. cosa = cos@ sin yw. Based on the far-field assumption in Section 2.2, we have where b is the distance between the two microphones, i.e. the length of the segment LR.
Fig.5 The signals after taking the Discrete Fourier Transform (DFT) of the noised signal d(t) with the source located at 9 = 50° and 6 = 85°, respectively. The two big peaks in the upper figure occur at +277/5 rad/sec (i.e., the angular velocity of the rotation of the microphone array) when 6 = 50°, whereas small peaks are present in the bottom figure when 6 = 85°
Fig.5 The signals after taking the Discrete Fourier Transform (DFT) of the noised signal d(t) with the source located at 9 = 50° and 6 = 85°, respectively. The two big peaks in the upper figure occur at +277/5 rad/sec (i.e., the angular velocity of the rotation of the microphone array) when 6 = 50°, whereas small peaks are present in the bottom figure when 6 = 85°
Estimated amplitude A, of the signal d(t) ‘using DFT for a sound source at 0 = 85° Estimated amplitude A, of the signal d(t) using DFT for a sound source at @ = 50°
Estimated amplitude A, of the signal d(t) ‘using DFT for a sound source at 0 = 85° Estimated amplitude A, of the signal d(t) using DFT for a sound source at @ = 50°
Fig. 7 Comparison of azimuth angle estimations using the 2D and 3D localization models when a sound source is located at 9 = O° and 6 = 20°, respectively
Fig. 7 Comparison of azimuth angle estimations using the 2D and 3D localization models when a sound source is located at 9 = O° and 6 = 20°, respectively
Fig. 8 RMSE between 2D and 3D localization model’s estimated azimuth angles
Fig. 8 RMSE between 2D and 3D localization model’s estimated azimuth angles
Fig. 9 Approximation of the elevation angle from the RMSE data using the least square fitted polynomial
Fig. 9 Approximation of the elevation angle from the RMSE data using the least square fitted polynomial
Fig.11 3D view of the system for distance localization been obtained by using Algorithm 1 and the microphone array has been regulated facing toward the sound source, as shown in Fig. 11. The proposed distance localization approach requires the microphone array, LR, to translate with a distance Ad along the line perpendicular to the center-source vector (on the horizontal plane). This translation shifts the center of the microphone array, O, toa new point, O’, and y is defined as the angle between vectors approach requires the microphone array, LR, to translate
Fig.11 3D view of the system for distance localization been obtained by using Algorithm 1 and the microphone array has been regulated facing toward the sound source, as shown in Fig. 11. The proposed distance localization approach requires the microphone array, LR, to translate with a distance Ad along the line perpendicular to the center-source vector (on the horizontal plane). This translation shifts the center of the microphone array, O, toa new point, O’, and y is defined as the angle between vectors approach requires the microphone array, LR, to translate
O'S and OS, as shown in Fig. 12. Note that the center of the robot, O, is unchanged. The objective is to estimate distance D between the center of the robot O and the source S.
O'S and OS, as shown in Fig. 12. Note that the center of the robot, O, is unchanged. The objective is to estimate distance D between the center of the robot O and the source S.
in Table 1. The complete EKF-based SSL procedure is illustrated in Fig. 13.
in Table 1. The complete EKF-based SSL procedure is illustrated in Fig. 13.
Table 1 EKF parameters In this section, we present the simulation results of the proposed localization technique for both angle and distance localization of a sound source.
Table 1 EKF parameters In this section, we present the simulation results of the proposed localization technique for both angle and distance localization of a sound source.
Table 2 Simulated room specifications
Table 2 Simulated room specifications
Fig. 13 Block diagram showing the process for the proposed complete angular and distance localization of a sound source using successive rotational and translational motions of a set of two microphones
Fig. 13 Block diagram showing the process for the proposed complete angular and distance localization of a sound source using successive rotational and translational motions of a set of two microphones
Fig. 15 Sound source locations with a fixed distance of 5 m to the center of the robot in the simulated room using the EKF with parameters given in Table 1. Four different sets of experiments were performed keeping the source at different locations. In first two sets of experiments, the source was placed in all four quadrants including the axes at different distances, keeping the elevation constant at 20° and 60°. To validate the performance of the proposed solution to the non-observability conditions, other two sets of experiments were performed by keeping the sound source at elevation close to 0° and 90°. The results of the localization are presented in Tables 3 and 4. It can be seen that orientation localization is achieved with errors less than 4° using speech as well as white noise sound source. Large errors are observed when the elevation of the sound source is around 0° and 90°. Further, the errors with source elevation around 0° is less as compared to source elevation around 90°. This was achieved by using polynomial curve fitting approach mentioned in Section 4.2, with RMS E;hreshold = 1.9°, which corresponds to 6 = 15° on the fitted curve shown in Fig. 9 . The value djjpesnoid was calculated as 0.022 m (which corresponds to 6 = 85°, thereby giving an accuracy of 5° when the sound source gets close to 90° elevation) for the simulated environment with specification given in Table 2.
Fig. 15 Sound source locations with a fixed distance of 5 m to the center of the robot in the simulated room using the EKF with parameters given in Table 1. Four different sets of experiments were performed keeping the source at different locations. In first two sets of experiments, the source was placed in all four quadrants including the axes at different distances, keeping the elevation constant at 20° and 60°. To validate the performance of the proposed solution to the non-observability conditions, other two sets of experiments were performed by keeping the sound source at elevation close to 0° and 90°. The results of the localization are presented in Tables 3 and 4. It can be seen that orientation localization is achieved with errors less than 4° using speech as well as white noise sound source. Large errors are observed when the elevation of the sound source is around 0° and 90°. Further, the errors with source elevation around 0° is less as compared to source elevation around 90°. This was achieved by using polynomial curve fitting approach mentioned in Section 4.2, with RMS E;hreshold = 1.9°, which corresponds to 6 = 15° on the fitted curve shown in Fig. 9 . The value djjpesnoid was calculated as 0.022 m (which corresponds to 6 = 85°, thereby giving an accuracy of 5° when the sound source gets close to 90° elevation) for the simulated environment with specification given in Table 2.
7.5 Simulation Results for Distance Localization Speech and white-noise sounds were also used to test the performance of the distance localization. A single sound source was placed at different locations and the ITD signal was recorded while the microphone array was continuously shifted for 200 steps each with a distance of Ad = 0.00043 m. The results are summarized in Tables 5 and 6. The key parameters of the EKF are given in Table 1. The results for the distance localization with a sound source placed at different locations are shown in Fig. 19. It is observed that the error in the estimation converges quickly and a total shift of microphone array of approximately 3 cm is sufficient for the estimates to completely converge to and remain in the three standard deviation bounds. The average of absoute error in the estimation is found to be less than 0.6 m in both the case of speech as well as white noise sound sources.
7.5 Simulation Results for Distance Localization Speech and white-noise sounds were also used to test the performance of the distance localization. A single sound source was placed at different locations and the ITD signal was recorded while the microphone array was continuously shifted for 200 steps each with a distance of Ad = 0.00043 m. The results are summarized in Tables 5 and 6. The key parameters of the EKF are given in Table 1. The results for the distance localization with a sound source placed at different locations are shown in Fig. 19. It is observed that the error in the estimation converges quickly and a total shift of microphone array of approximately 3 cm is sufficient for the estimates to completely converge to and remain in the three standard deviation bounds. The average of absoute error in the estimation is found to be less than 0.6 m in both the case of speech as well as white noise sound sources.
Fig. 17 Average of absolute errors in azimuth estimation using the 3D localization model without using the proposed methodology. Relatively large errors illustrate the non-observability condition in azimuth angle estimation with sound source placed around 90° elevation, as described by Theorem 2 The digitally generated audio signals using a MATLAB program and three 12-channel Digital-to-Analog converters
Fig. 17 Average of absolute errors in azimuth estimation using the 3D localization model without using the proposed methodology. Relatively large errors illustrate the non-observability condition in azimuth angle estimation with sound source placed around 90° elevation, as described by Theorem 2 The digitally generated audio signals using a MATLAB program and three 12-channel Digital-to-Analog converters
running at 44,100 cycles each second per channel were amplified using AudioSource AMP 1200 amplifiers before they were played one at a time from an array of 36 loudspeakers. Twenty-four loudspeakers were evenly arranged in the horizontal plane circling around the robot. The loudspeakers and the robot’s pinnae were at the same height. The second layer of 8 loudspeakers (45° spacing) was on the sphere at approximately 30° in elevation, and another layer of 4 loudspeakers (90° spacing) at approximately 60° in elevation. All loudspeakers were 1.5 m away from the listening position. The two microphones separated with a 0.18 m distance were installed on the KEMAR dummy head temporarily mounted on a rotating chair which was rotated at an approximate rate of 32°/s for about two circles in the middle of the room. The value D/b > 2.7 which satisfies the far-field approximation [13]. The data collected in the second rotation was used for the EKF. Motion data was collected by the gyroscope mounted on the top of the dummy head. The audio signals were amplified and collected by a sound card which were then stored on a desktop computer for further processing. The ITD was processed with a generalized cross-correlation model [29] in each time frame corresponding to the 120 Hz sampling rate of the gyroscope. The computation was completed by a MATLAB program on a desktop computer. Raw data with a single sound source located at three different locations were collected.
running at 44,100 cycles each second per channel were amplified using AudioSource AMP 1200 amplifiers before they were played one at a time from an array of 36 loudspeakers. Twenty-four loudspeakers were evenly arranged in the horizontal plane circling around the robot. The loudspeakers and the robot’s pinnae were at the same height. The second layer of 8 loudspeakers (45° spacing) was on the sphere at approximately 30° in elevation, and another layer of 4 loudspeakers (90° spacing) at approximately 60° in elevation. All loudspeakers were 1.5 m away from the listening position. The two microphones separated with a 0.18 m distance were installed on the KEMAR dummy head temporarily mounted on a rotating chair which was rotated at an approximate rate of 32°/s for about two circles in the middle of the room. The value D/b > 2.7 which satisfies the far-field approximation [13]. The data collected in the second rotation was used for the EKF. Motion data was collected by the gyroscope mounted on the top of the dummy head. The audio signals were amplified and collected by a sound card which were then stored on a desktop computer for further processing. The ITD was processed with a generalized cross-correlation model [29] in each time frame corresponding to the 120 Hz sampling rate of the gyroscope. The computation was completed by a MATLAB program on a desktop computer. Raw data with a single sound source located at three different locations were collected.
algorithm shown in Fig. 10. Table 7 shows the estimation results obtained using the 3D localization model. It can be seen that the RMSE of the difference between the estimated azimuth values using respectively the 2D and 3D models works well in checking the zero elevation condition. The right two subfigures in Fig. 21 are generated when the actual elevation angle is 30°. It can be seen that the azimuth estimations using the 2D and 3D localization models are obviously different while the elevation estimation using the 3D model is fairly accurate, which verifies the proposed
algorithm shown in Fig. 10. Table 7 shows the estimation results obtained using the 3D localization model. It can be seen that the RMSE of the difference between the estimated azimuth values using respectively the 2D and 3D models works well in checking the zero elevation condition. The right two subfigures in Fig. 21 are generated when the actual elevation angle is 30°. It can be seen that the azimuth estimations using the 2D and 3D localization models are obviously different while the elevation estimation using the 3D model is fairly accurate, which verifies the proposed
of 44,100 Hz for recording the sound signal coming from the sound source. Flex adapters were used to hold the microphones. WALLEYE microphone evaluation board assembly was used for data acquisition. The angular speed of the rotation of the microphone array was controlled by a bipolar stepper motor with gear ratio adjusted to 0.9° per
of 44,100 Hz for recording the sound signal coming from the sound source. Flex adapters were used to hold the microphones. WALLEYE microphone evaluation board assembly was used for data acquisition. The angular speed of the rotation of the microphone array was controlled by a bipolar stepper motor with gear ratio adjusted to 0.9° per
thereby satisfying the far-field approximation [13]. The room size was 4.5 m x 6 m x 3 m with the RT60 value as 670 ms. The value djhreshold Was calculated as 0.026 m for the robotic platform which corresponds to 6 = 85° as shown in the Fig. 23. The estimation results are shown in Fig. 24 and Table 8. step and rotating at an angular velocity of 27/5 rad/sec. which was controlled by an Arduino microprocessor. A slip ring was used to avoid the twisting of the wires. The distance between two microphones was kept constant as 0.3 m. An audio (music) was played in a loud speaker which was used as a sound source kept at different locations with D > 1.5m audio (music) was played in a loud speaker which was used
thereby satisfying the far-field approximation [13]. The room size was 4.5 m x 6 m x 3 m with the RT60 value as 670 ms. The value djhreshold Was calculated as 0.026 m for the robotic platform which corresponds to 6 = 85° as shown in the Fig. 23. The estimation results are shown in Fig. 24 and Table 8. step and rotating at an angular velocity of 27/5 rad/sec. which was controlled by an Arduino microprocessor. A slip ring was used to avoid the twisting of the wires. The distance between two microphones was kept constant as 0.3 m. An audio (music) was played in a loud speaker which was used as a sound source kept at different locations with D > 1.5m audio (music) was played in a loud speaker which was used
Table 7 Experimental results using KEMAR dummy head: Orienta- tion localization using the 3D model
Table 7 Experimental results using KEMAR dummy head: Orienta- tion localization using the 3D model
Fig. 20 Setup of the KEMAR dummy head on a rotating chair in the middle of the sound treated room [45]
Fig. 20 Setup of the KEMAR dummy head on a rotating chair in the middle of the sound treated room [45]
Fig. 22 A two-microphone system equipped on a ground robot
Fig. 22 A two-microphone system equipped on a ground robot
(RMSE: difference between azimuth estimations using the 2D and 3D models, respectively)
(RMSE: difference between azimuth estimations using the 2D and 3D models, respectively)
Fig.23 The signal after taking the Discrete Fourier transform (DFT) of the noised signal d(t) with the source located at 6 = 86° and at a distance of 2m from the center of the robotic platform, showing small peaks in |X y(q@)| and the maximum of Aa is less than the threshold of 0.026 m in the bottom figure
Fig.23 The signal after taking the Discrete Fourier transform (DFT) of the noised signal d(t) with the source located at 6 = 86° and at a distance of 2m from the center of the robotic platform, showing small peaks in |X y(q@)| and the maximum of Aa is less than the threshold of 0.026 m in the bottom figure
Related topics:
Sound Source LocalizationInteraural Time DifferenceMicrophone ArraysIndoor LocalizationAcoustic source localization and TrackingSound Localization
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved