JP2007078545A - Object detection system and voice conference system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an object detection system which detects the position and direction of an object more accurately and will not limit the detection range of the object. <P>SOLUTION: The voice conference system 1 is provided with functions for inputting a voice signal to a speaker array 2 and inputting the voice signal from a microphone array 3, and comprises a searching signal generation section 26 for generating a search voice signal and inputting it into the speaker array 2; first and second beam adjustment sections 13, 23 for adjusting the focuses of a voice beam and a sound collecting beam; a directivity control section 161 for controlling to make the focuses of these beams overlap each other; a timer 163 for measuring the search time for the voice to be output from the speaker array 2, reflecting at the focus position, and input into the microphone array 3; a signal processing section 25 for searching the voice signal from the microphone array 3 whether a component of the searching voice signal is included therein during the search time; and a position detection section 162 for deciding that there is an object in the path of the voice beam, when the component of the search voice signal is included. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an object detection apparatus and an audio conference apparatus that detect an object using an audio beam from a speaker array and a collected sound beam from a microphone array.

  2. Description of the Related Art Conventionally, there is known an object detection device that collects sound output from a speaker with a microphone and detects an object such as an obstacle or a person using the collected sound signal. For example, Patent Literature 1 describes a driving support device that detects an obstacle located around a vehicle and presents it to a driver. In this driving support apparatus, a transmitter (non-directional speaker) transmits an audible signal wave at a wide angle. The signal wave reflected by the obstacle is detected by a receiver array (microphone array) in which a plurality of receivers (microphones) are arranged.

  Phase correction is performed on the signal wave detected by each microphone of the microphone array. That is, the driving support apparatus divides the detection area into a plurality of areas, and stores each delay time until the sound from the division area is input to each microphone according to each division area. And a driving assistance apparatus performs phase correction by providing the delay time according to the division area corresponding to the signal wave detected with each microphone. This phase correction is performed for all the divided areas.

  The driving support device combines the signals after phase correction by superimposing them. As a result, only the sound components from the corresponding segmented regions are intensified by matching the phases of the signals. On the other hand, other audio components are weakened because the phases are not matched. For this reason, a sound collecting beam (having directivity) that collects sound specialized to the sound from the corresponding segmented region is formed, and thereby, it is possible to collect sound specialized to the sound from the segmented region.

The driving support device collects the sound of each divided area using the sound collection beam of each divided area and generates an intensity distribution indicating the intensity of each collected sound. Then, the driving support device detects that there is an obstacle at a position where the intensity is strong in the intensity distribution. In this way, the conventional driving support apparatus can detect the presence of an obstacle and the position of the obstacle.
JP-A-10-20034

  In the conventional object detection device, a signal wave in the audible band is transmitted at a wide angle using an omnidirectional speaker. Then, by collecting the sound of each divided area with the sound collecting beam, the intensity of the detected sound of each divided area is acquired, and it is detected that an object such as an obstacle is located in the divided area having a high intensity. For this reason, the sound reflected by the obstacle wraps around the other partitioned area where the obstacle exists and becomes noise. Obstacles may not be detected accurately due to such multipath that collects noise from other areas to be searched.

  Accordingly, an object of the present invention is to provide an object detection apparatus and an audio conference apparatus that can detect an object more accurately in order to solve the above problems.

  In order to solve the above problems, the present invention employs the following means.

  (1) The present invention has a function of inputting an audio signal to a speaker array and inputting an audio signal from a microphone array, generating a search audio signal and inputting the audio signal to the speaker array, and the speaker array A first beam adjusting unit for focusing the sound beam from the sound source, a second beam adjusting unit for focusing the sound collecting beam of the microphone array, and the focus of the sound beam and the sound collecting beam are overlapped with each other. A directivity control unit that controls the focusing of the first and second beam adjustment units, and a time measuring unit that measures the search time from when the sound is output from the speaker array until it is reflected at the focal position and input to the microphone array A search unit for searching whether the audio signal input from the microphone array includes a component of the search audio signal at the search time; and An object detection apparatus comprising: a determination unit configured to determine that there is an object in a path of an audio beam from the speaker array to a focus when it is detected that a component of a voice signal is included. is there.

  According to the configuration of the present invention, the search sound signal is generated and input to the speaker array by the signal generator. The first beam adjustment unit focuses the sound of the sound signal for search from the speaker array (sound for search). At the same time, the second beam adjustment unit performs focusing of the collected sound beam from the microphone array. Here, the focal points of the sound beam and the sound collecting beam are controlled by the directivity control unit so as to overlap.

  When the object is located in the output direction of the sound beam, the sound beam is reflected by the object. If the object is near the focal position, the reflected sound is collected by the sound collecting beam. The search unit searches for whether or not the audio signal input from the microphone array includes the component of the search audio signal. Then, when the search unit detects that the component of the audio signal for search is included, it is determined that there is an object in the path of the audio beam from the speaker array to the focal point.

  As described above, since the sound beam is output using the speaker array instead of the omnidirectional speaker as in the prior art, the sound is not output to the region (direction) outside the search range, and the reflected sound outside the search range is It is possible to effectively prevent multipaths that wrap around.

  In addition, the search time until the sound output from the speaker array is reflected at the focal position and input to the microphone array is measured by the timer unit. Then, the search unit searches for whether or not the audio signal input from the microphone array includes the component of the search audio signal only during the time measurement. Accordingly, it is possible to prevent the reflected sound reflected at the position beyond the focal position from being received by the microphone array and erroneously detecting the object.

  (2) In the above object detection apparatus according to the present invention, the determination unit uses the output timing of the sound beam from the speaker array, the input timing of the component of the sound signal for exploration to the microphone array, and the output direction of the sound beam. Then, the position of the object is detected.

  According to this configuration, it is possible to measure the distance of the object from the speaker array or the my array using the sound speed and the output timing and input timing. Further, the position direction of the object can be measured by the output direction of the sound beam. For this reason, it is possible to detect the position of the object from the distance and the position direction of the object.

  As described above in (1), it is possible to measure the position direction of the object more accurately as compared with the conventional technique using an omnidirectional speaker. In addition, since the distance of the object can be accurately measured, the detection position of the object can be measured more accurately than before.

  (3) In the object detection device according to the present invention, the signal generation unit generates an aperiodic pulse train as a search audio signal. This facilitates the determination unit to acquire the input timing of the component of the search audio signal. That is, when a periodic pulse train is used as the search audio signal, the pulse shape from a position shifted by a predetermined cycle (for example, one cycle) becomes the same as the pulse shape from the start position. For this reason, erroneous detection of the input start timing of the exploration audio signal from the microphone array is likely to be performed.

  On the other hand, when a non-periodic pulse train is used as a sound signal for exploration, the pulse shape from the shifted position will be different from the pulse shape from the start position. However, this is easier than using a periodic pulse train search audio signal.

  (4) The present invention is the audio conference device using the object detection device according to any one of claims 1 to 3 in the object detection device, wherein the signal generation unit can be used as an audio signal for exploration. An audio conference apparatus that generates an audio signal composed of a frequency component of listening sound, and wherein the determination unit detects a conference attendee as the object.

  According to the configuration of the present invention, the signal generation unit generates an audio signal composed of the frequency component of the audible sound as the search audio signal. The position and direction of the attendees of the conference are detected by the search voice. Audio conferencing devices typically have a configuration for outputting audible sounds. In the present invention, since an audio signal composed of frequency components of audible sound is used, it is possible to output search sound using a configuration that outputs audible sound that is normally provided in an audio conference apparatus. In addition, the audible sound is weak in straightness, but is output as a sound beam by the speaker array, so that this weakness in straightness can be compensated.

  According to the present invention, since the sound beam is output using the speaker array, it is possible to effectively prevent multipath that does not output sound outside the search range and collects reflected sound outside the search range. it can. Also, whether the audio signal input from the microphone array contains the component of the audio signal for exploration only during the time measured until the audio signal output from the speaker array is reflected at the focal position and input to the microphone array. Explored. As a result, it is possible to effectively prevent multipath in which reflected sound reflected at a position beyond the focal position is received by the microphone array, and to prevent erroneous detection of an object.

  As described above, multipath of reflected sound in other areas (directions) of the search range can be effectively prevented, and multipath of reflected sound at positions beyond the focal position can be effectively prevented. Therefore, the accuracy of detection of the object can be improved.

  With reference to FIGS. 1-6, the audio conference apparatus which is embodiment of this invention is demonstrated in detail. The audio conference apparatus 1 transmits / receives a call signal to / from other (the other party) audio conference apparatus 1 (denoted as the other party apparatus 1 'for distinction) at a remote place, thereby the present audio conference apparatus. This is a device for conducting an audio conference between one user (conference attendee h) and a partner speaker using the partner device 1 '.

  FIG. 1 is a diagram showing the appearance of the audio conference apparatus 1 as viewed from above, the propagation of audio for audio conference, and the sound collection range. In this figure, the audio conference apparatus 1 is arranged at a height position in the vicinity of the head of the seated conference attendee h by being arranged on the conference desk. In this figure, the front side of the audio conference apparatus 1 is described as -Y side, the rear side as Y side, the right side as X side, and the left side as -X side.

  The audio conference apparatus 1 includes a long casing 1A having a substantially rectangular parallelepiped shape, and includes a speaker array 2 on the upper side of the casing 1A on the -Y side. Note that the speaker array 2 is built in the housing 1A and cannot be visually seen from the outside, but in FIG. Similarly, for convenience of explanation, the speaker array 2 is shown on the Y side instead of the -Y side.

  The speaker array 2 includes eight speaker units SP (SP1 to SP8) arranged in a line shape in the longitudinal direction. Each speaker unit SP is arranged so that the sound emission surface is located on the -Y side. When an audio signal is input to the speaker units SP1 to SP8, an audio beam is output from the speaker array 2 so as to be directed in the -Y direction. The The sound beam from the speaker array 2 contains the voice of the other party's caller received from the other party device 1 '.

  The directivity (directivity direction and directivity range) of the sound beam from the speaker array 2 can be controlled by the delay time added to the sound signal input to each speaker unit SP. That is, a delay time (time indicated by a bold arrow in the figure) such that the sound from each speaker unit SP reaches the focal point P at the same timing is added to each signal input to each speaker unit SP. Thus, focusing can be performed so that the sound beam is directed to the focal point P.

  In this way, by directing the sound beam from the speaker array 2 in a narrow directivity range including the conference attendees h (h1 to h3), the conference audio can be provided only to the conference attendees h. . In other words, it is possible to effectively prevent the sound leakage of the conference audio for other people in the conference attendee h, and even if an audio conference is performed with a remote place in a normal office room instead of a conference room, Does not interfere with the work of others in the room meeting attendees.

  Further, a microphone array 3 is disposed on the lower stage of the housing 1A on the -Y side, that is, on the lower stage of the speaker array 2. Note that since the microphone array 3 is built in the housing 1A, it cannot be visually seen from the outside, but in FIG. In addition, although the speaker array 2 is positioned so as to overlap the upper side of the microphone array 3, the speaker array 2 and the microphone array 3 are shown in parallel in the horizontal direction for convenience of explanation.

  The microphone array 3 includes eight microphones M (M1 to M8) arranged in a line shape in the longitudinal direction. Each microphone M is arranged so that the sound collection side faces the -Y side. The microphone array 3 can collect the sound specialized to the sound at the search position by the sound collection beam having directivity to the predetermined search position. That is, the audio conference apparatus 1 stores each delay time until the sound from the search position reaches each microphone M, and performs phase adjustment with the corresponding delay time for the audio signal collected by each microphone M. As a result of this phase adjustment, the phases of the audio signals collected by the microphones M are matched with respect to the audio component from the search position.

  On the other hand, for audio components from positions away from the search position, the phases of the audio signals collected by the respective microphones M after phase adjustment become inconsistent. For this reason, by combining the audio signals collected by the microphones M after phase adjustment, the audio components from the search position are strengthened, and the audio components originating from a position away from the search position are weakened. As a result, the microphone array 3 can collect the sound specialized to the sound at the search position.

  In the audio conference apparatus 1, the sound collecting beam is focused on the position of the conference attendee h. As a result, it is possible to collect only the voice of the conference attendee h, and effectively prevent the conference attendee h from being difficult to hear by collecting noises such as environmental sounds. Is done.

  Further, when there are a plurality of meeting attendees h, for example, when there are three meeting attendees h1 to h3 as shown in the figure, the positions of these three meeting attendees h1 to h3 (positions P1 to P3). Focus the sound collecting beam. Then, the audio conference apparatus 1 compares the strengths of the collected voices at the positions P1 to P3, and adds and synthesizes the weights of the voices at the positions P1 to P3 according to the strength.

  The voice signal after the addition synthesis is output from the audio conference apparatus 1 to the counterpart apparatus 1 ′. As a result, a call voice specialized to the voice of the conference attendee h can be transmitted to the counterpart apparatus 1 ′. In this way, it is possible to effectively prevent the sound of the conference attendee h from being difficult to hear due to the noise component, and it is possible to suitably hold the audio conference between the audio conference device 1 and the counterpart device 1 ′. .

  As described above, in the present embodiment, when a voice conference is performed with the counterpart device 1 ′ using the voice conference device 1, the voice is set so that the focal point P is focused on the position of the conference attendee h from the speaker array 2. A beam is output. In addition, the focused sound beam is focused on the meeting attendee h. For this reason, the audio conference apparatus 1 performs a process (position detection process) for detecting the position of the meeting participant h (target object) in advance before the audio conference before the audio conference. The audio conference apparatus 1 performs focusing of the audio beam and the sound collection beam using the position of the conference attendee h acquired by the position detection process.

  Hereinafter, a position detection process executed by the audio conference apparatus 1 will be described with reference to FIG. FIG. 2 is a diagram showing the propagation range of the sound beam and the sound collection range of the sound collection beam when the position detection process is executed. The audio conference apparatus 1 outputs the search audio from the speaker array 2 as an audio beam. Here, in the position detection process, focusing is performed so that the position of the focal point P of the sound beam from the speaker array 2 and the focal position of the sound collection beam of the microphone array 3 overlap.

  As a result, when there is a meeting attendee h in the output direction of the sound beam, the sound beam is reflected by the meeting attendee h and collected by the sound collecting beam. Using the search sound output timing from the speaker array 2 and the search sound input timing from the microphone array 3, the distance L1 from the conference attendee h to the front side surface of the audio conference apparatus 1 according to the following equation (1). Can be measured.

Distance L1 = time t1 / 2 × C (1)
The time t1 is the time from the search sound output timing to the search sound input timing by the microphone array 3, and C is the speed of sound.

  Then, the position of the meeting attendee h can be detected using the output direction of the sound beam and the distance L1. In this way, the speaker array 2 outputs a sound beam, and the reflected sound of the sound beam is collected by the sound collecting beam, thereby detecting the position of the conference attendee h. For this reason, it is possible to effectively prevent multipath that does not output sound outside the propagation range (search range) of the sound beam and collects reflected sound outside the search range.

  In the position detection process, the audio conference apparatus 1 calculates the search time t2 until the reflected sound reflected by the focal point P of the audio beam is collected using the following equation (2). Then, it is determined whether or not the sound component for search is included in the input signal from the microphone array 3 only within the search time t2. As a result, the sound beam propagated further forward (in the −Y direction in the figure) after being focused at the focal point P is reflected to other objects and persons in the conference attendee h, and a multipath where the reflected sound is collected is obtained. Can be prevented.

t2 = L2 / C × 2 Formula (2)
L2 is the distance from the position of the microphone M farthest from the focal point P to the focal point P among the microphones M. When the microphone M farthest from the focal point P is used as a reference, if the reflected sound from the focal point P is not input by the microphone M farthest from the focal point P, the phase of each input signal from each microphone M is adjusted. This is because a sound collecting beam having directivity at the focal point P cannot be formed by the addition synthesis.

  As described above, the detection range of the conference attendee h is limited to the area between the audio conference apparatus 1 and the focal point P. Of course, even in the region between the audio conference apparatus 1 and the focal point P, sound is collected by the sound collecting beam, so that it becomes difficult to collect sound as the distance from the focal point P increases. Therefore, this is an effective position detection method when the position of the meeting attendee h is in the vicinity of the focal point P.

  The position detection method as described above is executed while moving the position of the focal point P within the search area (an area where a conference attendee may be seated). As a result, the position of the conference attendee can be detected in the search area. In the figure, the focal point P is moved so as to slide in the left-right direction while maintaining the position in the depth direction, but the position in the depth direction may be moved. As described above, the farther away from the focal point P, the less likely it is to collect sound, so this configuration makes it possible to detect the position of the meeting participant h more accurately.

  FIG. 3 is a block diagram schematically showing the configuration of the audio conference apparatus 1 shown in FIG. In addition to speaker units SP1 to SP8, input / output interface 11, echo canceller 12, delay unit 13, D / A (digital / analog) converter 14, amplifier 15, and control unit 16.

  The input / output interface 11 transmits / receives a digital audio signal to / from a counterpart apparatus 1 ′ connected to the communication cable via a communication cable (not shown) connected to the connection terminal 17. The echo canceller 12 receives a call voice signal (received voice signal) received from the counterpart apparatus 1 ′ via the input / output interface 11. The echo canceller 12 uses this input signal to generate a pseudo signal that simulates an echo component output from the speaker array 2 and fed back to the microphone array 3. The echo canceller 12 removes an echo component by removing a pseudo signal from an audio signal (described later) input from the microphone array 3.

  The delay unit 13 corresponds to the first beam adjustment unit of the present invention, and is provided by the number of speaker units SP1 to SP8 (eight). Hereinafter, when distinguishing each delay part 13, the number similar to what is corresponded among speaker units SP1-SP8 is attached as a subscript. For example, the delay unit 13 corresponding to the speaker unit SP1 is referred to as a delay unit 13-1. Each of the delay units 13-1 to 13-8 inputs the received audio signal to the D / A converter 14.

  Each of the delay units 13-1 to 13-8 receives the received audio signal from the echo canceller 12. Delay times D1 to D8 are set in the delay units 13-1 to 13-8, respectively. The delay units 13-1 to 13-8 delay the input received audio signal by delay times D1 to D8, and input them to the corresponding amplifiers 15 via the D / A converter 14 to thereby focus the audio beam. Align. In other words, the delay units 13-1 to 13-8 perform focusing so that the focal point P of the audio beam becomes the position of the conference attendee h based on the values of the delay times D1 to D8 added to the input received audio signal.

  The principle that the directivity of the sound beam is controlled by the values of the delay times D1 to D8 will be described. Sound is output from each speaker unit so as to propagate radially (circular). When the delay times D1 to D8 are set to the same time and the sounds from the speaker units are simultaneously output, only the components propagating parallel to each other are in phase and intensified. The components propagating in the other direction interfere with each other from the adjacent speaker units SP and are canceled out. Thereby, the synthesized speech from the speaker units SP1 to SP8 becomes an audio beam. The sound beam is directed in the front direction, and the beam width at the focal point P is substantially the same as the distance between the speaker unit SP1 and the speaker unit SP8.

  On the other hand, the delay times D1 to D8 are not set to the same time, but the delay times (for example, bold letters in FIG. 1) for the sound from the speaker units SP to reach the focal point P simultaneously and with a predetermined beam width. By setting the delay times D1 to D8 in (delay time indicated by an arrow), the focus P can be controlled.

  There are as many D / A converters 14 as the number of delay units 13. These D / A converters 14 (14-1 to 14-8) receive the received audio signal to which the delay time is given from the corresponding delay unit 13. The D / A converters 14-1 to 14-8 convert the received reception audio signal from a digital signal to an analog signal and input it to the corresponding amplifier 15.

  The amplifier 15 amplifies the signal level of the input audio signal. The amplifier 15 is provided in the number corresponding to the speaker units SP1 to SP8. Hereinafter, when distinguishing each amplifier 15, the number similar to what is corresponded among speaker units SP1-SP8 is attached as a subscript.

  The amplifier 15-1 to amplifier 15-8 receives the received audio signal from the delay units 13-1 to 13-8 via the D / A converter 14 (14-1 to 14-8). The amplifiers 15-1 to 15-8 amplify the signal level of the input reception audio signal and input the amplified signal levels to the corresponding speaker units SP1 to SP8. As a result, the voice of the received voice signal is emitted from the speaker units SP1 to SP8, and the voice of the other party from the other party apparatus 1 'is emitted.

  The control unit 16 includes a storage unit such as a CPU (Central Processing Unit) and a memory, a user interface such as an operation unit, and the like. By executing the program stored in the memory, the control unit 16 controls the operation of each unit of the audio conference apparatus 1 such as a call with the audio conference apparatus 1 ′. For example, the control unit 16 detects the position of the meeting attendee h by executing a position detection process. Then, the control unit 16 calculates delay times D1 to D8 such that the sound beam is focused on the detected position, and sets the delay times 13-1 to 13-8.

  In addition to the microphone array 3 and the above-described configuration, the audio conference device 1 is configured to output the audio signal of the conference attendee h to the counterpart device 1 ′, in addition to the microphone amplifier 21, the A / D conversion unit 22, the collection device. A sound beam forming unit 23, a band pass filter 24 and a signal processing unit 25 are provided.

  The microphone amplifier 21 receives each signal of the sound collected by each microphone M of the microphone array 3 from the microphone array 3 and amplifies each input signal. The A / D conversion unit 22 converts each amplified analog signal input from the microphone amplifier 21 into a digital signal. The A / D conversion unit 22 inputs the converted digital signals to the sound collection beam forming unit 23.

  The sound collection beam forming unit 23 corresponds to the second beam adjustment unit of the present invention, and includes a phase correction unit 231 and an addition unit 232. The phase correction unit 231 adjusts the phase of each digital signal input from the A / D conversion unit 22, and the addition unit 232 adds and synthesizes each digital signal after phase correction. As described above, this phase adjustment is performed using the delay times D11 to D18 until the sound from the search position reaches each microphone M. As a result, the phase of the audio component from the search position in the input signal can be matched, and the phases of the other components can be made inconsistent.

  For this reason, in each digital signal after addition synthesis, the level of the sound component from the search position is strengthened, and the levels of the other components are weakened. As a result, a sound collecting beam is formed that can collect the sound specifically for the sound from the search position. This search position is the position of the meeting attendee h as described above.

  The sound collecting beam forming unit 23 is provided with a plurality of channels (here, three channels of the sound collecting beam forming units 23A to 23C), and each channel 23A to 23C receives a sound collecting beam focused on the positions of the conference attendees h1 to h3. Form. The audio signals after the addition and synthesis by the channels 23A to 23C are input to the band pass filter 24, respectively.

  The phase correction unit 231 is realized by, for example, a delay time buffer memory (not shown) such as a shift register. The delay time buffer memory stores each digital signal input from the A / D conversion unit 22 markedly, and the stored value is read by being delayed by the phase correction and input to the addition unit 232. . As a result, phase correction can be performed with a predetermined delay time. The phase correction is controlled by the control unit 16. That is, the control unit 16 detects the position of the conference attendee h by a position detection process described later, and controls the phase correction so as to form a sound collecting beam having the detected position as a focal position.

  The band-pass filter 24 performs a convolution operation on the input audio signal with a filter coefficient for cutting other frequency bands of human speech, and a signal processing unit 25 (corresponding to the search unit of the present invention). To enter. As a result, only the frequency band components of human speech are extracted and input to the signal processing unit 25.

  Audio signals collected specifically for the positions of the conference attendees h1 to h3 are input to the signal processing unit 25 from the channels 23A to 23C via the bandpass filter 24. The signal processing unit 25 compares the levels of the input audio signals, and adds and synthesizes the audio signals with weighting according to the level ratio. The signal processing unit 25 inputs the audio signal (transmission audio signal) after the addition synthesis to the echo canceller 12. As a result, the audio signal at the position of the conference attendee h1 who is speaking can be reflected in the transmitted audio signal. The signal processing unit 25 is realized by a DSP, for example.

  In the echo canceller 12, as described above, the echo component is removed by removing the pseudo signal from the call voice signal input from the signal processing unit 25. The call voice signal after the removal of the echo component is transmitted to the counterpart apparatus 1 ′ via the input / output interface 11 and the connection terminal 17. As a result, the voice collected by the voice conference apparatus 1 ′ is input to the partner apparatus 1 ′, and a voice conference can be performed with the partner party.

  Furthermore, the audio conference apparatus 1 includes a search signal generation unit 26 (corresponding to the signal generation unit of the present invention), and executes position detection processing using the configuration of the search signal generation unit 26 and each unit described above. The function of each part in the position detection process will be described below.

  The control unit 16 functions as a directivity control unit 161, a position detection unit 162, a timer 163, and a position storage unit 164 by executing the stored program.

  The directivity control unit 161 calculates delay times D1 to D8 set in the delay units 13-1 to 13-8 so that the focal point of the sound beam becomes a predetermined search position, and delay units 13-1 to 13-. Set to 8. At the same time, the directivity control unit 161 calculates the delay times D11 to D18 set in the phase correction unit 231 so as to focus the collected sound beam at a position overlapping the focal position of the sound beam, and sets it in the phase correction unit 231. . The delay times D11 to D18 need only be set for one channel among the sound collection beam forming units 23A to 23C.

  The position detection unit 162 corresponds to the determination unit of the present invention, instructs the search signal generation unit 26 to generate a search audio signal, and activates a timer 163 (corresponding to the time measurement unit of the present invention) to measure time. Let it begin. The position detector 162 outputs a sound beam from the speaker array 2 (generates a sound signal for exploration), reflects it at the focal position, and calculates the time until the sound is collected by the microphone array 3, that is, the exploration time t2 described above. To do. Then, the position detection unit 162 causes the signal processing unit 25 to detect the search audio component only during the period from the start of timing until the search time t2 elapses. As a result, as described above, the sound for exploration reflected to the object or person positioned in the depth direction from the focal point is input to the microphone array 3 and is effectively prevented from being erroneously detected when the conference attendee h is present. .

  In addition, the position detection unit 162 receives a notification of detection start timing when the signal processing unit 25 detects a sound component for search. The position detection unit 162 sets the time from the generation of the search audio signal to the notification of the detection of the search audio component as the time t1, and uses the above-described equation (1) to call the audio conference device 1 from the conference attendee h. The distance L1 to the front side surface is calculated.

  In calculating the distance L1, the time is negligible so that the time can be ignored. Therefore, the time from the generation of the search audio signal to the input to the speaker array 2 and the search audio are input to the microphone array 3. Thus, the time until the search sound component is detected by the signal processing unit 25 is ignored. However, if the time is too long to be ignored, the distance L1 may be calculated after correcting for this time.

  The position detection unit 162 detects the position of the meeting attendee h using the calculated distance L1 and the output direction of the audio beam. The position detection unit 162 stores the detected position in the position storage unit 164. After the end of the position detection process, the directivity control unit 161 calculates delay times D1 to D8 and D11 to D18 that focus the sound beam and the sound collection beam on the detected position. The phase correction unit 231 is set. As a result, the sound beam is focused on the position of the conference attendee h1, the sound collection beam is focused, and the sound beam can be output with the directivity at the position of the conference attendee h1, and the conference attendee h1. It is possible to collect sound specialized for the sound from the position. For this reason, a voice conference can be suitably performed with the other party using the other party apparatus 1 '.

  The search signal generation unit 26 generates a search audio signal according to an instruction from the position detection unit 162 and inputs it to the delay units 13-1 to 13-8. This search audio signal is composed of frequency components of audible sound. As a result, the sound for exploration can be output using the configuration of the speaker array 2 and the D / A converter 14 that are normally provided for the audio conference apparatus 1 to hold an audio conference. Although the audible sound has poor straightness, this poor straightness can be compensated by being output as a sound beam by the speaker array 2.

  The search audio signal is composed of an aperiodic pulse train. An example of this non-periodic pulse train generation method will be described below. FIG. 4 is a diagram for explaining a method for generating an audio signal for search. First, the search signal generator 26 generates a digital audio signal (carrier signal) having a continuous waveform as shown by the waveform a1. The search signal generator 26 generates a pseudo-random number sequence b1 together with the carrier wave signal. The pseudo-random number sequence b1 is a binary random number using, for example, an M sequence or a Gold code.

  Then, the search signal generator 26 turns on / off (amplitude modulation) the carrier signal with the pseudo-random number sequence b1 to generate the search audio signal c1. Specifically, the search signal generator 26 generates a pseudo random number when the clock timing arrives. The clock cycle is, for example, one second interval. In the figure, the position of the bold arrow indicates the clock timing. Usually, the detection time accuracy of the input timing of the search signal to the microphone array 3 is determined by the clock cycle and is about 1/10 of the clock cycle.

  That is, the search signal generator 26 generates one of two values “0” or “1” as a pseudo-random number. When “0” is generated, the level of the carrier signal a1 is modulated to 0, and when “1” is generated, the level of the carrier signal a1 is not changed. As a result, the search audio signal c1 is generated. In the example shown in the figure, the pseudo-random number sequence b1 is “001101011”, and the search audio signal c1 is a pulse train composed of pulses in the period “1”. Note that the frequency of the search audio signal is, for example, 900 to 7 kHz. Of course, the preferred frequency varies depending on the width between the speaker units SP and the overall length of the speaker array 2. In addition, the preferred length of the time length varies depending on the sequence length of the pseudo-random numbers, has a length that can prevent false detection and accuracy of detection timing, and is long enough to keep the search time as short as possible. Preferably, it is not too much.

  The search audio signal generated by the search signal generation unit 26 is input to the delay unit 13. The delay unit 13 is provided with delay times D1 to D8 for focusing the sound beam at the search position, and the sound signal with the delay time is input to the speaker array 2 via the D / A converter 14-amplifier 15. The

  The audio beam output from the speaker array 2 is reflected by the conference attendee h and collected by the microphone array 3 when the conference attendee h is on the path of the audio beam. Each signal collected by each microphone M is input to the sound collection beam forming unit 23 via the microphone amplifier 21 -A / D conversion unit 22. Each signal is input to a channel in which delay times D11 to D18 are set. The sound collection beam forming unit 23 performs phase adjustment of each signal input in the set delay times D11 to D18, and adds and synthesizes each signal after phase adjustment, thereby specializing in the sound from the focal point P. To form a sound collecting beam. The sound collection beam forming unit 23 inputs the audio signal after the addition synthesis to the band pass filter 24.

  The band pass filter 24 is set with a filter coefficient for cutting off other components in the frequency band of the search voice. The band pass filter 24 performs a convolution operation on the input audio signal with the filter coefficient and inputs the result to the signal processing unit 25. As a result, only the frequency band component of the search voice can be extracted from the input voice signal and input to the signal processing unit 25.

  The signal processing unit 25 searches for whether or not the sound signal for search is included in the sound signal input from the band pass filter 24. As described above, since a non-periodic pulse train is used as the search audio signal, the input timing of the search audio to the microphone array 3 can be easily and accurately detected. FIG. 5 is a diagram illustrating a method for detecting the input timing of the search sound.

  The signal processing unit 25 detects the sound component for search by comparing the pulse waveform a2 of the sound signal for search with the waveform b2 of the input sound signal. That is, the control unit 16 is notified of the pulse waveform a2 of the search audio signal from the search signal generation unit 26, and the control unit 16 notifies the signal processing unit 25 of this pulse waveform a2. The signal processing unit 25 compares the notified pulse waveform a2 with the pulse waveform b2 of the audio signal input from the bandpass filter 24, and when both pulse waveforms a2 and b2 match, the search audio component Is determined to have been detected.

  FIG. 5A is a diagram showing a state in which the pulse waveform a2 and the pulse waveform b2 are compared when the search audio signal is a periodic pulse train. FIG. 4B is a diagram showing a state in which the pulse waveform a2 and the pulse waveform b2 are compared in the case of the present embodiment, which is an aperiodic pulse train. Note that the position of the dotted line is the clock timing, and the comparison of waveform matching is started from each clock timing.

  When the pulse train is periodic, as shown in (a), the shape of both pulse waveforms a2 and b2 is as shown in (a) at the clock timing T2 at a time deviated from the clock timing T1 at the start of the search audio component. It does not match. However, at the clock timing T3 shifted by one cycle, both pulse waveforms a2 and b2 coincide. For this reason, it is difficult to determine whether the start position of the search audio component is the clock timing T1 or the clock timing T3.

  On the other hand, when the search audio signal is composed of an aperiodic pulse train, only the pulse waveform b2 from the start time T1 of the search audio component matches the pulse waveform a2, as shown in (b), The pulse waveform b2 and the pulse waveform a2 from the time of deviation do not match. For this reason, it is easy to detect the start time of the sound component for search, and it is possible to accurately detect the time of disclosure.

  The signal processing unit 25 notifies the control unit 16 of the waveform matching start time as the search voice input timing. As described above, the control unit 16 calculates the position of the meeting attendee h using the notified search voice input timing and search voice output timing.

  FIG. 6 is a flowchart showing position detection processing executed by the audio conference apparatus 1 shown in FIG. This position detection process is executed when the user instructs execution of the position detection process using the operation unit constituting the control unit 16. First, the directivity control unit 161 calculates the delay times D1 to D8 described above and sets the delay times 13-1 to 13-8 to focus the sound beam (S1). At the same time, the directivity control unit 161 calculates the delay times D11 to D18 described above and sets the phase correction unit 231 to focus the sound collection beam (S2).

  Thereafter, the position detection unit 162 instructs the search signal generation unit 26 to generate a search audio signal (S3), and starts timing using the timer 163 (S4). The search signal generator 26 generates a search audio signal and inputs it to the speaker array 2 via the delay unit 13 -D / A converter 14 -amplifier 15. The position detection unit 162 causes the signal processing unit 25 to start detecting a sound component for search (S5).

  The position detection unit 162 determines whether or not the input timing of the search sound is notified from the signal processing unit 25 (S6). When the exploration audio input timing is notified (YES in S6), the position detection unit 162 uses the input timing and the exploration audio output timing (the occurrence timing of the exploration audio signal) to perform a meeting by the method described above. The position of the attendee h1 is calculated and stored in the position storage unit 164 (S7).

  Thereafter, the position detection unit 162 causes the directivity control unit 161 to determine whether the focal positions of the sound beam and the sound collection beam have been moved in all directions of the search area (S8). If the directivity control unit 161 determines that the focal position has been moved in all directions (YES in S8), the position detection unit 162 ends this process. On the other hand, when the directivity control unit 161 determines that the focus position has not been moved in all directions (NO in S8), the position detection unit 162 returns the process to step S1, and the focus is different in step S1. It is changed to the position.

  On the other hand, when the position detection unit 162 determines that the input timing of the search sound is not notified from the signal processing unit 25 (NO in S6), the position detection unit 162 determines whether the search time t2 has elapsed (S9). . If it is not determined that the exploration time t2 has elapsed (NO in S9), the position detection unit 162 returns this processing to step S6, and if it is determined that the exploration time t2 has elapsed (YES in S9). The position detection unit 162 executes step S8 described above.

  According to the position detection process described above, in this embodiment, the directivity control unit 161 controls the focus of the sound collecting beam of the microphone array 3 and the focus of the sound beam of the speaker array 2 to overlap. Then, an exploration sound beam is output from the speaker array 2, and when the conference attendee h is on the sound beam path, the exploration sound reflected by the conference attendant h is collected by the microphone array 3. The The position of the meeting attendee h is detected by the position detection unit 162 based on the input and output timing of the search sound and the output direction of the sound beam. As a result, the position of the meeting attendee h can be detected more accurately than in the prior art.

  That is, unlike the conventional technique in which sound for exploration is output at a wide angle using an omnidirectional speaker, a sound beam is output using the speaker array 2. Therefore, it is possible to effectively prevent multipath that does not output sound outside the search range and collects reflected sound outside the search range. Further, the audio signal input from the microphone array 3 includes the component of the audio signal for exploration only during the time until the audio signal output from the speaker array 2 is reflected at the focal position and input to the microphone array 3. It will be explored. Accordingly, it is possible to prevent the reflected sound reflected at the position beyond the focal position from being received by the microphone array (multipath) and erroneously detecting the attendee h.

  As described above, multipath of reflected sound in other areas (directions) of the search range can be effectively prevented, and multipath of reflected sound at positions beyond the focal position can be effectively prevented. Therefore, it is possible to improve the accuracy of position detection of the meeting attendee h.

  The present embodiment can employ the following modified examples.

  (1) In the present embodiment, the number of speaker units SP is eight. However, the number is not limited to this, and it is sufficient that at least the number capable of controlling the directivity and beam width of the sound beam is provided.

  (2) In the present embodiment, the object detection device of the present invention is applied to an audio conference device. However, the present invention is not limited to this, and the present invention is applied to any device having a function of detecting an object. Can do. For example, the present invention may be applied to a driving support device that detects an obstacle located around the vehicle and presents it to the driver.

  (3) In this embodiment, the speaker array 2 in which the speaker units SP are arranged in a line and the microphone array 3 in which the microphones M are arranged in a line are used. However, the present invention uses the speaker array 2 and the microphone array 3. It is not limited to the configuration. For example, a speaker array 2 or a microphone array 3 including speaker units SP and microphones M arranged in a matrix shape, a honeycomb shape, a circular shape, or the like may be used. In this case, the position of the object in the height direction can also be detected.

It is a figure which shows the propagation and sound collection range of the audio | voice conference audio | voice when the audio conference apparatus 1 was seen from the upper direction. FIG. 2 is a diagram showing the propagation range of the sound beam and the sound collection range of the sound collection beam when the position detection process is executed. It is a block diagram which shows roughly the structure of the audio conference apparatus 1 shown in FIG. It is a figure for demonstrating the production | generation method of the audio | voice signal for a search. It is a figure which shows the detection method of the input timing of the audio | voice for search. It is a flowchart which shows the position detection process which the audio conference apparatus shown in FIG. 3 performs.

Explanation of symbols

1-voice conference device 1'-partner device 2-speaker array 3-microphone array 13-delay unit (first beam adjustment unit) 16-control unit 161-directivity control unit 162-position detection unit (determination unit) 163- Timer (timer) 23 (23A to 23C) -sound collecting beam forming unit (second beam adjusting unit) 25-signal processing unit (searching unit) 26-searching signal generating unit (signal generating unit) c1-searching sound Signal h (h1-h3)-Conference attendee P-Focus

Claims (4)

A function to input audio signals from the microphone array and input audio signals to the speaker array,
A signal generation unit for generating a search sound signal and inputting the sound signal to the speaker array;
A first beam adjustment unit for focusing an audio beam from the speaker array;
A second beam adjustment unit for focusing the sound collection beam of the microphone array;
A directivity control unit that controls the focusing of the first and second beam adjustment units so that the focus of the sound beam and the focus of the sound collection beam overlap.
A timekeeping unit that measures the exploration time from when the sound is output from the speaker array until it is reflected at the focal position and input to the microphone array;
An exploration unit for exploring whether the audio signal input from the microphone array includes the component of the audio signal for exploration at the exploration time;
A determination unit that determines that there is an object in the path of the sound beam from the speaker array to the focal point when the search unit detects that the component of the search sound signal is included;
An object detection apparatus comprising: The determination unit detects the position of the object using the output timing of the sound beam from the speaker array, the input timing of the component of the sound signal for exploration to the microphone array, and the output direction of the sound beam.
The object detection apparatus according to claim 1.   The object detection device according to claim 2, wherein the signal generation unit generates a non-periodic pulse train as an audio signal for search. An audio conference device using the object detection device according to claim 1,
The signal generation unit generates an audio signal composed of frequency components of audible sound as an audio signal for exploration,
The determination unit detects meeting attendees as the object.
An audio conference apparatus.

Images