This application claims the benefit of Taiwan application Serial No. 96111886, filed Apr. 3, 2007, the subject matter of which is incorporated herein by reference.
1. Field of the Invention
The invention relates in general to an air vibration signal localization system and an air vibration signal localization method thereof, and more particularly to a sound source localization system and a sound source localization method.
2. Description of the Related Art
A conventional method adopts an image localization system to locate a target. The image localization system includes an image acquiring device and an image processing unit. The image acquiring device acquires an image of the target to be located, and thus outputs image data to the image processing unit. The image processing unit calculates the location of the target according to the image data outputted from the image acquiring device.
However, the data quantity of an image signal is significant large, the processing speed of the image processing unit is relative slow. Furthermore, the hardware requirement of the image localization system is high, thereby increasing the manufacturing cost and the power consumption. In addition, the image localization system tends to be restricted by the environmental brightness or the weather condition. For example, when the electrical power is off or the heavy fog is generated, the image localization system cannot locate the target. So, it is an important subject of the invention to provide a different localization system to improve the drawback caused in the conventional image localization system.
The invention is directed to a sound source localization system and a sound source localization method, in which time domain signals outputted from sound capturing devices, such as microphones, are respectively transformed into frequency domain signals, and then a cross spectrum process is performed according to the frequency domain signals to determine time differences of arrival (TDOA), and the sound source is located according to the time differences of arrival and locations of the sound capturing devices. Consequently, the processing speed of the localization system can be increased, and the hardware requirement of the power consumption of the localization system can be reduced. Furthermore, the sound source localization system is not restricted by the environmental brightness or the weather condition, and can precisely locate the sound source even the electrical power is off or the heavy fog is generated.
According to a first aspect of the present invention, a sound source localization system is provided. The sound source localization system includes sound capturing devices and an arithmetic unit. The sound capturing devices respectively sense a sound source as the time domain signals. The arithmetic unit transforms the time domain signals into frequency domain signals, performs a cross spectrum process according to the frequency domain signals to determine time differences of arrival (TDOA), and locates the sound source according to the time differences of arrival and locations of the sound capturing devices.
According to a second aspect of the present invention, a sound source localization method is provided. The sound source localization method includes the following steps. First, sound capturing devices sense a sound source and output time domain signals correspondingly. Next, the time domain signals are respectively transformed into frequency domain signals. Then, a cross spectrum process is performed according to the frequency domain signals to determine time differences of arrival (TDOA). Finally, the sound source is located according to the time differences of arrival and locations of the sound capturing devices.
The invention will become apparent from the following detailed description of several embodiments. The following description is made with reference to the accompanying drawings.
Since the model of the plane sound field is only a special example in which the sound source S in the model of the spherical sound field is located at infinity, the model of the spherical sound field will be described in the following embodiment.
After receiving the time domain signals T(1) to T(n) outputted from the sound capturing devices 110(1) to 110(n), the arithmetic unit 120 transforms the time domain signals T(1) to T(n) into frequency domain signals by way of fast Fourier transform (FFT), for example.
The arithmetic unit 120 performs a cross spectrum process according to the frequency domain signals T(1) to T(n) to determine time differences of arrival (t2−t1) to (tn−t1) between the time instants when the wave front of the sound source S enters the sound capturing devices 110(2) to 110(n) and the time instant when the wave front of the sound source S enters the sound capturing device 110(1), and locates the sound source S according to the time differences of arrival (t2−t1) to (tn−t1), locations of the sound capturing devices 110(1) to 110(n) and the sound velocity c. The sound source localization system 10 may further include a temperature sensor 130 for outputting an environmental temperature to the arithmetic unit 120 so that the arithmetic unit 120 calibrates the sound velocity c according to the environmental temperature.
In detail, after the time domain signals T(1) to T(n) are transformed into the frequency domain signals, the arithmetic unit 120 samples the frequency domain signals in a specific frequency range to get a plurality of sampled frequency domain signals, and calculates phase differences between the sampled frequency domain signals to determine the time differences of arrival (t2−t1) to (tn−t1) between the time instants when the wave front of the sound source S enters the sound capturing devices 110(2) to 110(n) and the time instant when the wave front of the sound source S enters the sound capturing device 110(1). The specific frequency range may be in a set of predetermined values. Alternatively, the arithmetic unit 120 may find the specific frequency range which covers at least a part of the range of maximum power signals.
For example, the frequency of the human sound typically falls within the frequency band ranging from 200 Hz to 2 K Hz. The specific frequency range can be pre-set as ranging from 200 Hz to 2 K Hz so that the arithmetic unit 120 can analyze the talking sound of the human beings and the interference of the environmental noise can be avoided.
Alternatively, the arithmetic unit 120 finds the specific frequency range which covers the range of maximum power signals. Because the sound source S is typically the sound source with the strongest energy, the arithmetic unit 120 analyzes the specific frequency range corresponding to the maximum power among the frequency domain signals T(1) to T(n) to thus avoid the interference of the environmental noise.
For example, the coordinates of the sound source S, the sound capturing device 110(1), the sound capturing device 110(2) and the sound capturing device 110(3) are respectively (xs,ys), (x1,y1), (x2,y2) and (x3,y3). Substituting the coordinates into the distance formula can get the distances from the sound source S to the sound capturing devices 110(1) to 110(3) are
respectively. The distance difference b between the distance from the sound capturing device 110(2) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is
while the distance difference a between the distance from the sound capturing device 110(3) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is
Since the distance difference b between the distance from the sound capturing device 110(2) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is also equal to c(t2−t1), and the distance difference a between the distance from the sound capturing device 110(3) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is also equal to c(t3−t1) so that the following simultaneous equations (1) and (2) can be listed:
Because the coordinates (x1,y1), (x2,y2) and (x3,y3), the sound velocity c, the TDOA (t3−t1) and the TDOA (t2−t1) are known, the coordinates (xs,ys) can be obtained after the arithmetic unit 120 solves the simultaneous equations (1) and (2). The coordinates (xs,ys) represent the location of the sound source S.
In addition, the arithmetic unit 120 may also locate the sound source S according to the time differences of arrival, the locations of the sound capturing devices, the sound velocity c and the trigonometric function.
For example, the trigonometric function is a cosine formula. The distance from the sound source S to the sound capturing device 110(1) is r, the distance d1 from the sound capturing device 110(2) to the sound capturing device 110(1) is √{square root over ((x2−x1)2+(y2−y1)2)}{square root over ((x2−x1)2+(y2−y1)2)}, and the distance d2 from the sound capturing device 110(3) to the sound capturing device 110(1) is √{square root over ((x3−x1)2+(y3−y1)2)}{square root over ((x3−x1)2+(y3−y1)2)}.
Substituting the distance r, the distance d1, the distance d2, the distance difference and the distance difference b into the cosine formula can get the simultaneous equations (3) and (4):
Because the distance d1 is √{square root over ((x2−x1)2+(y2−y1)2)}{square root over ((x2−x1)2+(y2−y1)2)}, the distance d2 is √{square root over ((x3−x1)2+(y3−y1)2)}{square root over ((x3−x1)2+(y3−y1)2)}, the distance difference a is equal to c(t3−t1) and the distance difference b is equal to c(t2−t1), and the distances d1 and d2 and the distance differences a and b are all known, the distance r and cos θ1 can be obtained after the arithmetic unit 120 solves the simultaneous equations (3) and (4). The arithmetic unit 120 further takes an arc-cosine operation on the cos θ1 to get DOA θ1, and locates the sound source S according to the distance r and the DOA θ1.
For example, the coordinates of the sound source S and the sound capturing devices 110(1) to 110(4) are respectively (xs,ys,zs), (x1,y1,z1), (x2,y2,z2), (x3,y3,z3) and (x4,y4,z4). Substituting the coordinates into the distance formula can get the distances from the sound source S to the sound capturing devices 110(1) to 110(3), and the distances are respectively
Consequently, the distance difference between the distance from the sound capturing device 110(2) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is equal to
the distance difference between the distance from the sound capturing device 110(3) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is equal to
and the distance difference between the distance from the sound capturing device 110(4) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is equal to
The distance difference between the distance from the sound capturing device 110(2) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is also equal to c(t2−t1); the distance difference between the distance from the sound capturing device 110(3) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is also equal to c(t3−t1); and the distance difference between the distance from the sound capturing device 110(4) to the sound source S and the distance from the sound capturing device 110(1) to the sound source S is also equal to c(t4−t1). So, the simultaneous equations (1), (2) and (3) may be listed as follows:
The coordinates (x1,y1), (x2,y2), (x3,y3) and (x4,y4), the sound velocity c, the TDOA (t2−t1), the TDOA (t3−t1) and the TDOA (t4−t1) are known. So, the simultaneous equations (1) to (3) can be solved to obtain the coordinates (xs,ys,zs), which represent the location of the sound source S.
First, as shown in step 610, the sound capturing devices 110(1) to 110(n) are utilized to sense the sound source S, and output the time domain signals T(1) to T(n) to the arithmetic unit 120 correspondingly.
Next, as shown in step 620, the arithmetic unit 120 transforms the time domain signals T(1) to T(n) into the corresponding frequency domain signals. The arithmetic unit 120 transforms the time domain signals T(1) to T(n) into the frequency domain signals by Fourier transform (FT) or fast Fourier transform (FFT), for example.
Then, as shown in step 630, the arithmetic unit 120 performs the cross spectrum process according to the frequency domain signals to determine the time differences of arrival.
Finally, as shown in step 640, the arithmetic unit 120 locates the sound source S according to the time differences of arrival and the locations of the sound capturing devices 110(1) to 110(n).
The sound source localization system and its sound source localization method according to the embodiment of the invention can determine the time differences of arrival (t2−t1) to (tn−t1) through the cross spectrum process, and then locate the sound source according to the time differences of arrival and the locations of the sound capturing devices.
In addition, the sound source localization system and its sound source localization method according to the embodiment of the invention include the following features. First, the processing speed of the arithmetic unit is increased and the drawback of the too-slow processing speed of the conventional image localization system can be eliminated because the data quantity of the audio signal is smaller than that of the image signal. Furthermore, the manufacturing cost and the power consumption can be reduced because the hardware requirement of the sound source localization system is lower than that of the conventional image localization system. Moreover, the sound source localization system can precisely locate the sound source even if the electrical power is off or the heavy fog is generated because the sound source localization system is not restricted to the environmental brightness or the weather condition.
While the invention has been described by way of example and in terms of several embodiments, it is to be understood that the invention is not limited thereto. On the contrary, it is intended to cover various modifications and similar arrangements and procedures, and the scope of the appended claims therefore should be accorded the broadest interpretation so as to encompass all such modifications and similar arrangements and procedures.
Number | Date | Country | Kind |
---|---|---|---|
96111886 A | Apr 2007 | TW | national |
Number | Name | Date | Kind |
---|---|---|---|
4118599 | Iwahara et al. | Oct 1978 | A |
4208735 | Suzuki et al. | Jun 1980 | A |
4581758 | Coker et al. | Apr 1986 | A |
5060269 | Zinser | Oct 1991 | A |
5119424 | Asakawa et al. | Jun 1992 | A |
5222059 | Holman | Jun 1993 | A |
5365516 | Jandrell | Nov 1994 | A |
5465302 | Lazzari et al. | Nov 1995 | A |
5602926 | Ohashi et al. | Feb 1997 | A |
5778082 | Chu et al. | Jul 1998 | A |
6130949 | Aoki et al. | Oct 2000 | A |
6469732 | Chang et al. | Oct 2002 | B1 |
6593956 | Potts et al. | Jul 2003 | B1 |
6600824 | Matsuo et al. | Jul 2003 | B1 |
6618485 | Matsuo et al. | Sep 2003 | B1 |
6760449 | Matsuo et al. | Jul 2004 | B1 |
6774934 | Belt et al. | Aug 2004 | B1 |
6999593 | Rui et al. | Feb 2006 | B2 |
7039198 | Birchfield et al. | May 2006 | B2 |
7123727 | Elko et al. | Oct 2006 | B2 |
7127071 | Rui et al. | Oct 2006 | B2 |
20040240680 | Rui et al. | Dec 2004 | A1 |
Number | Date | Country |
---|---|---|
1560556 | Jan 2005 | CN |
1719277 | Jan 2006 | CN |
2000035474 | Feb 2000 | JP |
Number | Date | Country | |
---|---|---|---|
20080247566 A1 | Oct 2008 | US |