JP3976078B1 - Telephone device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inexpensive speech unit for preventing the occurrence of howling by canceling a sound component generated by a speaker from a transmission signal over a wide frequency band. <P>SOLUTION: An A/D conversion circuit 31 A/D converts a sound signal from a microphone M2 after lapse of a delay time to a sound wave having a frequency f1 from a point of time when an A/D conversion circuit 30 A/D converts a sound signal from a microphone M1. Estimating circuits 322 to 32n, when a phase difference of frequency f2 to fn component to a frequency f1 component of a sound signal collected by the microphone M2 is defined as deviation time Th2 to Thn, estimates a sound signal generated by the microphone M2 at deviation time Th2 to Thn from a point of a time when the A/D conversion circuit 31 performs A/D conversion based on an output of the A/D conversion circuit 31 to reduce sound volume from a speaker SP through attenuation circuit 351 to 35n and calculation circuit 36a. <P>COPYRIGHT: (C)2008,JPO&amp;INPIT

Description

  The present invention relates to a call device.

  Conventionally, there is a communication device installed indoors with an interphone system or the like, which includes a speaker that outputs sound from a communication device installed in another place, a microphone that inputs sound transmitted to the other communication device, and the like. Yes.

  Since howling occurs when the sound generated from the speaker wraps around the microphone, various measures for preventing howling are taken. For example, a delay circuit that includes a speaker and a pair of microphones, delays the output of a microphone closer to the speaker by a delay time of sound waves corresponding to the difference in distance between the two microphones and the speaker, and both the microphone and the speaker. Adjusting the level corresponding to the difference in distance to match the output level of both microphones with respect to the sound from the speaker, and using both microphone outputs through the delay circuit and the level adjusting amplifier circuit as both inputs There has been proposed a communication device that is provided with a differential amplifier circuit and uses the output of the differential amplifier circuit as a transmission signal.

In this communication device, after picking up the sound from the speakers with both microphones, the delay and the level are adjusted, and the sound components from the speakers input to both microphones are canceled by the differential amplifier circuit. Only the components are removed (cancellation process) to prevent howling. (For example, refer to Patent Documents 1 and 2).
Japanese Patent Laid-Open No. 11-41342 Japanese Patent No. 3226121

  However, the sound emitted from the speaker is transmitted to the microphone and converted into an audio signal, but the transfer function of the sound wave transmitted from the speaker to the microphone has a frequency characteristic, and the phase of the sound collected by the microphone, The amplitude depends on the frequency. Therefore, in the conventional configurations such as Patent Documents 1 and 2, the sound component from the speaker can be canceled near a specific frequency, but the sound component from the speaker cannot be canceled over a wide frequency band. . There is also a demand for cost reduction.

  The present invention has been made in view of the above-mentioned reasons, and its object is to cancel a voice component emitted from a speaker from a transmission signal over a wide frequency band and to prevent a howling from occurring at a low cost. Is to provide.

  According to the first aspect of the present invention, the speaker that outputs the transmitted sound information, the first microphone that collects the sound and outputs the sound signal, and the distance from the speaker are arranged at a position farther than the first microphone. A second microphone that collects sound and outputs a sound signal; and a signal processing unit that processes and transmits each sound signal output from the first and second microphones, the signal processing unit The first A / D conversion means for A / D converting the audio signal output from the first microphone, and the first frequency corresponding to the difference between the distances between the first and second microphones and the speaker. The sound signal from the second microphone is obtained when the delay time elapses from the timing when the first A / D conversion means performs A / D conversion on the sound signal from the first microphone, with the sound wave transmission time as the delay time. A / D The phase difference of the second frequency component with respect to the first frequency component of the audio signal collected by one of the second A / D conversion means and the first and second microphones is defined as a shift time. Based on the output of the A / D conversion means for converting the audio signal output from one microphone into a digital signal, the one microphone is at a timing shifted from the timing at which the A / D conversion means performs A / D conversion. And a process for making the output levels of the first and second microphones corresponding to the sound from the speaker coincide with each frequency band including the first and second frequencies, respectively. Level adjusting means applied to at least one of the outputs of the one A / D converting means and the estimating means and the output of the other A / D converting means; , Characterized by comprising a calculating means for outputting a difference between the second microphone of the speech signal.

  According to the present invention, since the voice component emitted from the speaker is canceled from the transmission signal for each frequency band, howling can be prevented over a wide frequency band. Furthermore, since the audio signal having a frequency that is not subjected to A / D conversion is estimated by the estimation means, there is no need for an A / D conversion means or a large number of A / D conversion means that operate at high speed, cost reduction, circuit scale reduction, Miniaturization can be achieved. That is, it is possible to provide a low-cost communication device that can cancel the sound component emitted from the speaker from the transmitted signal over a wide frequency band and prevent the occurrence of howling.

  According to a second aspect of the present invention, in the first aspect, the estimating means uses the phase difference of the second frequency component with respect to the first frequency component of the sound signal collected by the second microphone as a shift time, Based on the output of the second A / D conversion means, the audio signal output from the second microphone is estimated at a timing shifted from the timing at which the second A / D conversion means performs A / D conversion. It is characterized by.

  According to the present invention, since the sound signal output from the second microphone is estimated by the estimation means, the second A / D conversion means or a large number of second A / D conversion means that operate at high speed is not necessary. Cost reduction, circuit scale reduction, and miniaturization can be achieved.

  According to a third aspect of the present invention, in the first aspect, the estimating means uses the phase difference of the second frequency component with respect to the first frequency component of the sound signal collected by the first microphone as a shift time, Estimating an audio signal output from the first microphone at a timing shifted from the timing at which the first A / D conversion unit performs A / D conversion based on the output of the first A / D conversion unit. It is characterized by.

  According to the present invention, since the sound signal output from the first microphone is estimated by the estimation means, the first A / D conversion means or a large number of first A / D conversion means that operate at high speed is not necessary. Cost reduction, circuit scale reduction, and miniaturization can be achieved.

  According to a fourth aspect of the present invention, in any one of the first to third aspects, the estimating means is configured such that an audio signal changes along a straight line connecting two digital values continuously output by one A / D conversion means. The estimation operation is performed based on the straight line.

  According to the present invention, an audio signal that has not been A / D converted can be estimated by a simple method.

  According to a fifth aspect of the present invention, in any one of the first to fourth aspects, the frequency band including the first frequency is passed from the output of the one A / D conversion means, and the second frequency is determined from the output of the estimation means. Including a first filter means for passing a frequency band including the second filter means for allowing the output of the other A / D conversion means to be separated and passed for each frequency band. The first and second microphones that have passed through the first and second filter means and the level adjusting means are added together with the difference for each frequency band and output.

  According to the present invention, since the voice component emitted from the speaker is canceled from the transmission signal after being divided for each frequency band by the filter means, occurrence of howling can be surely prevented.

  As described above, according to the present invention, there is an effect that it is possible to provide a low-cost communication device that can cancel the sound component emitted from the speaker from the transmission signal over a wide frequency band and prevent the occurrence of howling. is there.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(Embodiment 1)
The communication device A of the present embodiment is shown in FIGS. 2 to 4, and a housing A1 is configured by a body A10 having an opening formed on the rear surface and a cover A11 covering the opening of the body A10. A speaker SP, a microphone board MB1, a call switch SW1, and a voice processing unit 10 are provided.

  As shown in FIG. 4, the audio processing unit 10 is composed of an IC including a communication unit 10a, echo cancellation units 10b and 10c, an amplification unit 10d, and a signal processing unit 10e, and is arranged in the housing A1. A voice signal transmitted from the communication device A installed in another room or the like via the information line Ls is received by the communication unit 10a, amplified by the amplification unit 10d via the echo cancellation unit 10b, and then the speaker. Output from SP. Further, by operating the call switch SW1, a call can be made and each audio signal input from the microphone M1 (first microphone) and the microphone M2 (second microphone) on the microphone board MB1 is received by the signal processing unit 10e. After being subjected to signal processing to be described later, the signal passes through the echo cancel unit 10c, and is transmitted from the communication unit 10a to the communication device A installed in another room or the like via the information line Ls. That is, it functions as an intercom that allows two-way calls between rooms. Note that the power of the communication device A may be supplied from an outlet provided in the vicinity of the installation location or may be supplied via the information line Ls.

  As shown in FIG. 2, the speaker SP is a cylindrical yoke 20 formed of an iron-based material having a thickness of about 0.8 mm such as cold rolled steel plate (SPCC, SPCEN), electromagnetic soft iron (SUY), etc., and having one end opened. The circular support body 21 is extended from the opening end of the yoke 20 toward the outside.

  A cylindrical permanent magnet 22 (for example, residual magnetic flux density of 1.39 T to 1.43 T) formed of neodymium is disposed in the cylinder of the yoke 20, and the outer peripheral edge of the dome-shaped diaphragm 23 is a support. 21 is fixed to the edge surface.

  The diaphragm 23 is formed of a thermoplastic plastic (for example, a thickness of 12 μm to 50 μm) such as PET (PolyEthyleneTerephthalate) or PEI (Polyetherimide). A cylindrical bobbin 24 is fixed to the rear surface of the diaphragm 23, and is formed by winding a polyurethane copper wire (for example, φ0.05 mm) around a paper tube of kraft paper at the rear end of the bobbin 24. A voice coil 25 is provided. The bobbin 24 and the voice coil 25 are provided so that the voice coil 25 is positioned at the opening end of the yoke 20, and freely move in the front-rear direction in the vicinity of the opening end of the yoke 20.

  When an audio signal is input to the polyurethane copper wire of the voice coil 25, an electromagnetic force is generated in the voice coil 25 due to the current of the audio signal and the magnetic field of the permanent magnet 22, so that the bobbin 24 moves back and forth with the diaphragm 23. Visible in the direction. At this time, a sound corresponding to the audio signal is emitted from the diaphragm 23. That is, an electrodynamic speaker SP is configured.

  And the rib 11 is formed in the front inner side of housing A1 which the diaphragm 23 of speaker SP opposes, and the end surface of the convex part 21a protruded to the front side from the outer peripheral end part of the circular support body 21 of speaker SP. Comes into contact with the rib 11, and the speaker SP is fixed in a state where the diaphragm 23 faces the front surface of the housing A1 from the inside.

  When the speaker SP is fixed in the housing A1, the front air chamber Bf which is a space surrounded by the front inner side of the housing A1 and the front surface side (the diaphragm 23 side) of the speaker SP, the rear inner side and the inner side surface of the housing A1. And a rear air chamber Br which is a space surrounded by the back surface side (yoke 20 side) of the speaker SP. The front air chamber Bf communicates with the outside through a plurality of sound holes 12 provided on the front surface of the housing A1. The rear air chamber Br becomes a space that is insulated (not communicated) with the front air chamber Bf by closely contacting the end portion of the support 21 of the speaker SP and the rib 11 on the inner surface of the housing A1, and further covers the cover A11. Is in close contact with the rear opening of the body A10 to form a sealed space that is insulated from the outside.

  Next, as shown in FIG. 5, the microphone substrate MB1 has a pair of microphone bare chips BC1 and ICKa1 and a pair of microphone bare chips BC2 and ICKa2 mounted on one surface 2a of the module substrate 2, respectively, and bare chips BC1 and ICKa1. After connecting the wiring patterns (not shown) on the module substrate 2 and between the bare chips BC2, ICKa2 and the wiring patterns (not shown) on the module substrate 2 with wires W (wire bonding), A shield case SC1 is mounted so as to cover the pair of bare chips BC1 and ICKa1, and a shield case SC2 is mounted so as to cover the pair of bare chips BC2 and ICKa2, so that the microphone configured by the bare chips BC1, ICKa1, and shield case SC1 M1, bare chip BC2, I Ka2, and a composed microphone M2 with a shield case SC2.

  As shown in FIG. 6, in the bare chip BC (bare chip BC1 or BC2), an Si thin film 1d is formed on one surface side of the silicon substrate 1b so as to close the hole 1c formed in the silicon substrate 1b. An electrode 1f is formed between them via an air gap 1e, and a pad 1g for outputting an audio signal is further provided to constitute a capacitor type silicon microphone. Then, an external acoustic signal vibrates the Si thin film 1d, whereby the capacitance between the Si thin film 1d and the electrode 1f changes to change the amount of charge, and the pad 1g changes with this change in the amount of charge. , 1g, a current corresponding to the acoustic signal flows. In this bare chip BC, the silicon substrate 1b is die-bonded on the module substrate 2. In particular, the Si thin film 1d of the bare chip BC2 faces the sound hole F2 formed in the module substrate 2.

  The microphone M1 uses the bottom surface side of the shield case SC1 with the sound hole F1 as a sound collecting surface, and the microphone M2 uses the mounting surface side with respect to the module substrate 2 with the sound hole F2 as a sound collecting surface. The module substrate 2 has sound collecting surfaces in both directions opposite to each other. Since the microphone substrate MB1 configured in this manner has both the microphones M1 and M2 mounted on the one surface 2a of the module substrate 2, the thickness of the microphone substrate MB1 can be reduced.

  FIG. 7A is a plan view of the microphone board MB1 as viewed from the one surface 2a side of the module board 2. The module board 2 has a rectangular part 2f in which the microphone M1 is arranged and a rectangular part 2g in which the microphone M2 is arranged. And a connecting portion 2h that connects the rectangular portions 2f and 2g, and the rectangular portion 2g is formed larger than the rectangular portion 2f. A negative power supply pad P1, a positive power supply pad P2, an output 1 pad P3, and an output 2 pad P4 are provided along the edge of the rectangular portion 2g.

  As shown in FIG. 7B, the negative power supply pad P1 is connected to the negative side of the power supply voltage supplied from the outside, and the positive power supply pad P2 is connected to the positive side of the power supply voltage. Power is supplied to the microphones M1 and M2 through the pattern. Further, an audio signal collected by the microphone M1 is output from the output 1 pad P3 via a wiring pattern on the module substrate 2, and an audio signal collected by the microphone M2 is output from the output 2 pad P4. It is output via the upper wiring pattern. The ground of the audio signal output from the output pads P3 and P4 is shared by the negative power supply pad P1.

  Thus, the power of the microphones M1 and M2 is supplied from the common negative power supply pad P1 and the positive power supply pad P2, and the ground of each output of the microphones M1 and M2 is shared by the negative power supply pad P1, so that the number of pads The configuration can be simplified.

  Next, the operation of the microphone substrate MB1 will be described.

  First, each current flowing from the bare chips BC1 and BC2 according to the collected acoustic signal is impedance-converted by ICKa1 and Ka2 and converted into a voltage signal, and output from the output 1 pad P3 and the output 2 pad P4 as audio signals, respectively. Is done.

  The ICKa (ICKa1 or Ka2) has the circuit configuration of FIG. 8, and is a constant IC composed of a chip IC that converts the power supply voltage + V (for example, 5V) supplied from the power supply pads P1 and P2 into a constant voltage Vr (for example, 12V). A voltage circuit Kb is provided, a constant voltage Vr is applied to the series circuit of the resistor R11 and the bare chip BC, and a connection midpoint between the resistor R11 and the bare chip BC is connected to the junction type J-FET element S11 via the capacitor C11. Connected to the gate terminal. The drain terminal of the J-FET element S11 is connected to the operating power supply + V, and the source terminal is connected to the negative side of the power supply voltage via the resistor R12. Here, the J-FET element S11 is for electrical impedance conversion, and the voltage at the source terminal of the J-FET element S11 is output as an audio signal. Note that the ICKa impedance conversion circuit is not limited to the above-described configuration, and may be, for example, a circuit having a function of a source follower circuit using an operational amplifier, or an audio signal amplification circuit in the ICKa if necessary. May be provided.

  The microphone board MB1 can efficiently configure the signal lines and the power supply lines by performing signal transmission and power supply via the wiring pattern on the module board 2 as described above, and can be attached to the outer surface of the housing A1. Become. In this embodiment, one surface 2a of the module substrate 2 is arranged along the outer front surface of the housing A1, and the microphone M1 is inserted through the opening 13 on the front surface of the housing A1 so that the sound collection surface faces the front air chamber Bf, and the shield The sound hole F1 of the microphone M1 formed in the bottom surface of the case SC1 faces the diaphragm 23 of the speaker SP, and the sound emitted from the speaker SP can be reliably collected through the sound hole F1. The microphone M2 is fitted into a recess 14 provided in the front surface of the housing A1, and the sound hole F2 of the microphone M2 formed in the module substrate 2 is located outside (frontward) the housing A1 in the output direction of the speaker SP. Therefore, it is possible to reliably collect the sound transmitted from the speaker located in front of the communication device A and transmitted through the sound hole F2. If the distances from the center of the speaker SP to the centers of the microphones M1 and M2 are X1 and X2, respectively, X1 <X2.

  Further, the rear air chamber Br facing the back surface of the speaker SP is sealed in the housing A1, so that sound radiated from the back surface of the speaker SP is difficult to leak from the rear air chamber Br, and the speaker SP and the microphone M2 are not connected. Acoustic coupling is reduced. Furthermore, the sound radiated from the back surface of the speaker SP (the back surface of the diaphragm 23) has a phase reversed from that of the sound radiated from the surface of the speaker SP (the surface of the diaphragm 23). When the generated sound circulates forward, the sound radiated from the surface of the speaker SP cancels each other, the radiated sound pressure of the speaker SP decreases, and the speaker in front cannot hear the sound emitted by the speaker SP. However, since the sound radiated from the back surface of the speaker SP is difficult to leak to the outside of the housing A1 as described above, a decrease in the radiated sound pressure of the speaker SP due to the wraparound is prevented.

  Further, since the concave portion 14 in which the microphone M2 is accommodated is a separated space that does not communicate with the rear air chamber Br, the microphone M2 is more difficult to collect the sound emitted by the speaker SP, and the speaker SP and the microphone M2 are separated. The acoustic coupling is further reduced. That is, with the above configuration, the sound emitted from the speaker SP and the sound emitted from the speaker are separated and collected by the microphones M1 and M2.

  Further, when the microphone substrate MB1 is disposed in the housing A1, it is difficult to maintain the spatial insulation between the front air chamber Bf and the rear air chamber Br, but the microphone substrate MB1 is disposed in the housing as in the present embodiment. By attaching to the outer surface of A1, the spatial insulation between the front air chamber Bf and the rear air chamber Br can be maintained.

  And in this embodiment, in order to prevent the howling which generate | occur | produces when the microphones M1 and M2 pick up the audio | voice output of the speaker SP, it has the following structures.

  First, as shown in FIG. 1, the signal processing unit 10e accommodated in the audio processing unit 10 includes A / D conversion circuits 30 and 31, estimation circuits 322 to 32n, band pass filters 331 to 33n, It comprises pass filters 341 to 34n, attenuation circuits 351 to 35n, an arithmetic circuit 36a, and a timing control unit 37. Each operation of the A / D conversion circuits 30 and 31 is controlled by the timing control unit 37, and the operation of each circuit will be described below with reference to FIGS.

  First, the microphone M1 has a sound collection surface mounted toward the speaker SP, and the speaker SP is more effective than the sound (transmitted sound) emitted by the speaker H (see FIG. 1) located in front of the call device A. The sound that is emitted is collected with high sensitivity. On the other hand, the microphone M2 has a sound collection surface arranged forward, and collects sound emitted by the speaker H located in front of the communication device A with higher sensitivity than the sound emitted by the speaker SP.

  That is, the sound signal Y11 output from the microphone M1 has high sensitivity and large amplitude with respect to the sound emitted from the speaker SP, but has low sensitivity and small amplitude with respect to the sound emitted from the speaker H. The sound signal Y21 output from the microphone M2 has high sensitivity and large amplitude for the sound emitted by the speaker H, but has low sensitivity and small amplitude for the sound emitted from the speaker SP.

  Furthermore, at the time of transmission, both the sound emitted by the speaker H and the sound emitted by the speaker SP are collected by the microphones M1 and M2, but the distance X1 from the center of the speaker SP to the center of each of the microphones M1 and M2 , X2 satisfy X1 <X2, so that a phase difference occurs between the sound signal Y11 of the microphone M1 and the sound signal Y21 of the microphone M2 with respect to the sound from the speaker SP, and both the microphones M1, M2 and the speaker SP. The sound signal Y21 of the microphone M2 is compared with the sound signal Y11 of the microphone M1 by the sound wave delay time [Td = (X2-X1) / Vs] (Vs is the speed of sound) corresponding to the difference (X2-X1) in distance from the sound wave. Is delayed (see FIGS. 9A and 9B). The delay time Td is constant with respect to the frequency without depending on the frequency of the sound emitted from the speaker SP under ideal conditions (for example, a point sound source, a sealed housing structure, and no variation in CR of the circuit configuration). Although there is actually no ideal condition, it depends on the frequency of the sound emitted by the speaker SP.

  Similarly, the amplitudes of the audio signals Y11 and Y12 with respect to the sound from the speaker SP are also constant with respect to the frequency under the ideal conditions without depending on the frequency of the sound emitted from the speaker SP. Since it is not an ideal condition, it depends on the frequency of the sound emitted by the speaker SP.

  On the other hand, since the distances between the microphones M1 and M2 and the speaker H can be considered to be equal, the voice signals Y11 and Y21 of the microphones M1 and M2 have substantially the same phase with respect to the voice emitted by the speaker H. .

  The A / D conversion circuit 30 converts the analog audio signal Y11 of the microphone M1 into a digital signal, and the A / D conversion circuit 31 converts the analog audio signal Y21 of the microphone M2 into a digital signal. The A / D conversion operation of the A / D conversion circuit 30 is performed in synchronization with the rise of the control signal Si1 output from the timing control unit 37, and the A / D conversion operation of the A / D conversion circuit 31 is performed by the timing control unit. 37 is performed in synchronization with the rise of the control signal Si2, and the frequencies of the control signals Si1 and Si2 of this embodiment are set to 8 to 16 KHz, and the conversion cycle T1 = 62 in each A / D conversion circuit. .5 to 125 μsec. Further, since the phase of the audio signal Y21 of the microphone M2 is delayed by the delay time Td1 from the audio signal Y11 of the microphone M1 with respect to the audio of the frequency f1 (first frequency) emitted from the speaker SP, the clock signal By setting the operation of the timing control unit 37 in advance so that Si2 is generated with a delay time Td1 with respect to the clock signal Si1, the audio signals Y11 and Y21 output from the microphones M1 and M2 are the same at the frequency f1. A / D conversion is performed on the phase location (FIGS. 9C and 9D).

  Further, the A / D conversion circuits 30 and 31 perform A / D conversion by the ΣΔ method, and change the frequency transfer characteristic of only noise, which is a feature of the ΣΔ method, so that the noise component becomes a higher frequency region than the signal component. The distributed noise shaping enables A / D conversion with a high S / N ratio, and further, most of the processing is performed digitally, which facilitates the implementation of IC. Note that the resolution of A / D conversion is any one of 16 bits, 14 bits, and 12 bits.

  The A / D conversion circuit 30 outputs an audio signal Y12 composed of digital values a1, b1, c1, d1, e1,... A / D converted from the audio signal output from the microphone M1. (See FIG. 9 (e)), the A / D conversion circuit 31 is configured by the digital values a2, b2, c2, d2, e2,... Obtained by A / D converting the audio signal output from the microphone M2. The signal Y221 is output (see FIG. 9F), the digital value a1 of the microphone M1, the digital value a2 of the microphone M2, the digital value b1 of the microphone M1, the digital value b2 of the microphone M2, and the digital value c1 of the microphone M1. And the digital values c2,... Of the microphone M2 respectively correspond to each other so that the audio signals of the microphones M1, M2 have the same phase at the frequency f1. To have. However, as described above, the delay time Td depends on the frequency of the sound emitted from the speaker SP. In the configuration of this embodiment, the sound signals of the microphones M1 and M2 have the same phase at the frequency f1, but the frequency f1. The other phases do not have the same phase.

  Therefore, the following processing is performed with the phase differences of the components of the frequencies f2 to fn (second frequency) different from the component of the frequency f1 of the sound signal collected by the microphone M2 as the shift times Th2 to Thn. In the following processing, the digital values of the audio signals of the microphones M1 and M2 are stored in the memory and digitally processed. For the sake of explanation, the waveforms in FIGS. 12 to 14 are shown as analog waveforms. .

  First, the audio signal Y12 output from the A / D conversion circuit 30 and the Y221 output from the A / D conversion circuit 31 are branched into n systems, and the following processing is performed in each system. The audio signal Y12 output from the A / D conversion circuit 30 is separated from the component of the frequency band G1 including the frequency f1 by the bandpass filter 331, and is separated from the component of the frequency band G2 including the frequency f2 by the bandpass filter 332. ,..., Components of the frequency band Gn including the frequency fn are separated by the band pass filter 33n and distributed to n systems.

  Also, the audio signal Y221 output from the A / D conversion circuit 31 is input to the estimation circuits 322 to 32n in the n-1 system except for one system. The estimation circuits 322 to 32n linearly interpolate the digital values a2, b2, c2, d2, e2,... (Voice signal Y221) of the A / D conversion circuit 31, and the digital values a2, b2, c2, d2, e2 , ... Approximate that the distance is not a curve but a straight line. Based on this linear approximation, it is considered that the digital values a2, b2, c2, d2, e2,... Were output from the microphone M2 at the timing shifted by the shift times Th2 to Thn. Guess each audio signal. The shift times Th2 to Thn are times when the components of the frequencies f2 to fn included in the sound emitted from the speaker SP deviate from the components of the frequency f1 when the microphone M2 collects the sound emitted from the speaker SP. The components of the frequencies f2 to fn having the same phase as the component of the frequency f1 included in the audio signal Y221 are obtained by the estimation circuits 322 to 32n. That is, the estimation circuits 322 to 32n output the sound signal of the microphone M2 having the same phase as the sound signal Y12 of the microphone M1 at the frequencies f2 to fn. It should be noted that whether the estimated audio signal shifts in the direction of the earlier timing or the later timing of the digital values a2, b2, c2, d2, e2,. It depends on the height of ~ fn.

For example, as shown in FIG. 10, when the digital value a2 and b2 on the actual waveform Y21a of the audio signal Y21 of the microphone M2 is approximated by a straight line Y21b, the A / D conversion circuit 31 sets the digital value b2. The actual audio signal from the microphone M2 at the timing before the shift time Thm with respect to the sound wave of the frequency fm from the timing tb to be acquired is as on the waveform Y21a, but is assumed to be the audio signal ah on the straight line Y21b by linear interpolation. is there. Here, the audio signal from the speaker SP is to be expressed by exp (jωt), a2 = exp (jωt a) becomes, b2 = exp since the _{(jω (t a + T1)} ), inferred audio signal ah Becomes [Formula 1].

Note that the shift time Thm (Th2 to Thn) is set in advance by actual measurement for each frequency f2 to fn.

  [Equation 1] is summarized as [Equation 2].

  Therefore, the audio signal ah estimated by linear interpolation changes in amplitude and phase according to the term {(1+ (exp (−jωt) −1) · Thm / T1)}, and the amplitude characteristic Ah (ω) is expressed as 3], and the phase characteristic θh (ω) becomes [Formula 4].

  Further, ah, as, and b2 in FIG. 10 are represented by phasor vectors as shown in FIGS. 11 (a) and 11 (b). FIG. 11A shows a case where Thm <(T1 / 2). Note that linear interpolation can be performed in the same manner even if Thm <(T1 / 2), and FIG. 11B shows a phasor vector in the case of Thm> (T1 / 2).

  The estimation circuits 322 to 32n perform the estimation process of the audio signal of the microphone M2 using linear interpolation as described above, and the estimation circuit 322 is a microphone having the same phase as the audio signal Y12 of the microphone M1 at the frequency f2. The M2 audio signal Y222 is output, and the estimation circuit 32n outputs the microphone M2 audio signal Y22n having the same phase as the audio signal Y12 of the microphone M1 at the frequency fn. By performing linear interpolation in this way, it is possible to easily estimate the audio signal, and further, since it is not necessary to A / D convert the audio signal of the microphone M2 corresponding to the frequencies f2 to fn, the circuit scale can be reduced. Cost reduction and downsizing can be achieved.

  Next, the audio signal Y221 output from the A / D conversion circuit 31 passes the component of the frequency band G1 including the frequency f1 by the bandpass filter 341, and the audio signal Y222 output from the estimation circuit 322 is the bandpass filter. The component of the frequency band G2 including the frequency f2 passes through the filter 342, and the audio signal Y22n output from the estimation circuit 32n passes the component of the frequency band Gn including the frequency fn through the bandpass filter 34n.

  The audio signals Y131 and Y231 in the frequency band G1 output from the bandpass filters 331 and 341 are generated by operating the A / D conversion circuit 30 and the A / D conversion circuit 31 with the delay time Td1 shifted from each other. The phase is the same at f1. Further, the audio signals Y132, Y232,... Of the frequency band G2 output from the bandpass filters 332, 342, and the audio signals Y13n, Y23n of the frequency band Gn output from the bandpass filters 33n, 34n are the estimation circuits 322,. ..., 32n have the same phase at frequencies f2 to fn (see FIGS. 12A and 12B).

  The attenuation circuit 351 attenuates the audio signal Y131 of the microphone M1 that has passed through the bandpass filter 331 to generate an audio signal Y141, and the attenuation circuit 352 attenuates the audio signal Y132 of the microphone M1 that has passed through the bandpass filter 332 The attenuation circuit 35n generates an audio signal Y14n by attenuating the audio signal Y13n of the microphone M1 that has passed through the band-pass filter 33n, and generates the audio signal Y14n for each frequency band G1 to Gn. The level adjustment corresponding to the difference in distance between the microphones M1, M2 and the speaker SP (X2-X1) and the sensitivity difference between the microphones M1, M2 is performed, and the respective frequency bands G1 to Gn (frequency f2) of the sound emitted from the speaker SP. ˜fn), the output levels of both microphones M1 and M2 are matched (FIG. 1). (A) (b) reference).

  Next, the arithmetic circuit 36a subtracts the audio signal Y141 of the microphone M1 from the audio signal Y231 of the microphone M2, thereby generating an audio signal in which the audio component from the speaker SP is canceled in the frequency band G1, and the microphone M2 By subtracting the audio signal Y142 of the microphone M1 from the audio signal Y232, an audio signal in which the audio component from the speaker SP is canceled in the frequency band G2 is generated,..., ... from the audio signal Y23n of the microphone M2 By subtracting the audio signal Y14n, an audio signal in which the audio component from the speaker SP is canceled in the frequency band Gn is generated (see FIG. 14A), and the subtraction results in the frequency bands G1 to Gn are further obtained. All are added, and the sound component from the speaker SP is in each frequency band. Audio signal Ya that is canceled for each 1~Gn is generated (see FIG. 14 (b)). In FIG. 14A, the audio signal Y14n 'obtained by inverting the audio signal Y14n is added to the audio signal Y23n, so that the audio signal Y14n of the microphone M1 is subtracted from the audio signal Y23n.

  On the other hand, for the sound uttered by the speaker H in front of the microphones M1 and M2, the amplitude of the sound signal Y21 of the microphone M2 having the sound collection surface arranged toward the speaker H is such that the sound collection surface faces the speaker SP. It becomes larger than the amplitude of the audio signal Y11 of the arranged microphone M1. Further, since the signal from the microphone M1 is attenuated by the attenuation circuits 351 to 35n, the speech component from the speaker H included in the speech signals Y231 to 23n is the speech component from the speaker H included in the speech signals Y141 to 14n. Even bigger. That is, the amplitude difference between the speech component from the speaker H included in the speech signals Y141 to Y14n and the speech component from the speaker H included in the speech signals Y231 to Y23n increases, and the arithmetic circuit 36a performs the above subtraction process. Even if applied, the signal corresponding to the voice uttered by the speaker H remains in the voice signal Ya in a state where the amplitude is maintained sufficiently.

  In the audio signal Ya output from the signal processing unit 10e as described above, the audio component from the speaker SP is reduced for each of the frequency bands G1 to Gn, while the speaker H in front of the communication device A is directed to the microphones M1 and M2. In the audio signal Ya, the relative difference between the audio component from the speaker H to be kept and the audio component from the speaker SP to be reduced becomes large over a wide frequency band. The effect of preventing howling that occurs when the microphones M1 and M2 pick up the sound output of the speaker SP is improved.

  Using the signal processing unit 10e of the present embodiment, the frequency range G1 of 2 KHz or less optimized for the speaker sound cancellation amount at the frequency f1 = 1 KHz, and 2 KHz or higher optimized for the speaker sound cancellation amount at the frequency f2 = 3 KHz. The cancellation amount of the speaker sound when divided into two in the frequency band G2 is indicated by data Y1 in FIG. 15. When this data Y1 is approximated, it is represented by a curve Y2, and the cancellation amount is near both frequencies near 1 KHz and near 3 KHz. It has a peak characteristic and it can be seen that there is an effect of canceling speaker sound over a wide frequency band.

  Next, the effect of canceling speaker sound by the signal processing unit 10e of the present embodiment will be described using mathematical expressions. Assuming that the audio signal from the speaker SP is expressed by exp (jωt), the output Ya of the signal processing unit 10e is obtained when the phase of the audio signals of the microphones M1 and M2 is not matched and the difference between the two audio signals is taken. The remaining speaker sound is expressed by [Equation 5], and its amplitude characteristic U1 is expressed by [Equation 6].

  Next, when the estimation process using the linear interpolation of the present embodiment is performed, and the phase and amplitude of the audio signals of the microphones M1 and M2 are matched, the signal processing is performed when the difference between the two audio signals is taken. The speaker sound remaining in the output Ya of the unit 10e is expressed by [Expression 7], and the amplitude characteristic U2 is expressed by [Expression 8].

Note that the deviation time Thm and the A / D conversion cycle T1 are the same as those in the above [Equation 1] to [Equation 4], and the description thereof will be omitted.

[Amplitude characteristic | U1 | ² −amplitude characteristic | U2 | ² ] is expressed by [Equation 9]. However, D = Thm / T1.

The first term [2D (1-D) (1 + cos (ωT1))] on the right side of [Equation 9] is positive because 0 <ωT1 <π. The second term [2D {cos (ωT1 / 2) cos {ω ((T1 / 2) −Thm)}}] on the right side is 0 <ωT1 <π, and Thm <(T1 / 2). If there is, it becomes positive. Therefore, [| U1 | ² − | U2 | ² ] is positive, and it is more output when the phase adjustment is performed after the estimation process using the linear interpolation of the present embodiment is performed than when the phase adjustment is not performed. The speaker sound remaining in Ya has decreased and it can be said that there is a canceling effect. Although the case of Thm <(T1 / 2) has been described above, the same effect can be obtained even when Thm> (T1 / 2).

  Next, the audio signal output from the signal processing unit 10e is output to the echo cancellation unit 10c, and the echo cancellation units 10b and 10c (see FIG. 4) perform further processing to prevent further howling.

  First, the echo canceling unit 10c captures the output of the echo canceling unit 10b as a reference signal, and performs an operation on the output of the signal processing unit 10e, thereby further processing the audio signal that has circulated from the speaker SP to the microphones M1 and M2. Cancel. On the other hand, the echo canceling unit 10b also captures the output of the echo canceling unit 10c as a reference signal and performs an operation on the output of the communication unit 10a, thereby wrapping the audio signal from the speaker to the microphone on the other party side Cancel.

  Specifically, the echo cancellation units 10b and 10c are configured by a speaker SP-microphone M1, M2-signal processing unit 10e-echo cancellation unit 10c-communication unit 10a-echo cancellation unit 10b-amplification unit 10d-speaker SP. By adjusting the amount of loss in a variable loss means (not shown) provided in the loop circuit, the loop gain is set to 1 or less to prevent howling. Here, it is assumed that the smaller signal level of the transmission signal and the reception signal is not important, and the transmission loss of the variable loss circuit inserted in the transmission line having the smaller signal level is increased.

  Note that the number of microphones M1 and M2 is not limited to one each, and a plurality of microphones may be provided as the microphones M1 and M2.

(Embodiment 2)
In the signal processing unit 10e according to the first embodiment, the band-pass filters 331 to 33n and 341 to 34n are provided in the previous stage of the arithmetic circuit 36a to perform frequency division. However, in the present embodiment, the filter means is not provided. A signal processing unit 10e that can obtain substantially the same effect as the frequency division of 1 will be described. In addition, the same code | symbol is attached | subjected to the structure similar to Embodiment 1, and description is abbreviate | omitted.

  First, the signal processing unit 10e of the present embodiment includes an arithmetic circuit 36b as shown in FIG. 16, and the arithmetic circuit 36b is an attenuation circuit in which the output of the A / D conversion circuit 30 is directly input as an audio signal of the microphone M1. The audio signals Y141 to Y14n output from 351 to 35n are input, the audio signal Y221 output from the A / D conversion circuit 31 and the audio signals Y222 to Y22n output from the estimation circuits 322 to 32n as the audio signals of the microphone M2. Is entered directly. That is, as in the first embodiment, the sound signals of the microphones M1 and M2 for the sound emitted from the speaker SP are set to the same phase for each of the frequency bands G1 to Gn, and the sound signals of the microphones M1 and M2 are further attenuated by the attenuation circuits 351 to 35n. Each signal having the same amplitude for each of the frequency bands G1 to Gn is input to the arithmetic circuit 36b without passing through the bandpass filters 331 to 33n and 341 to 34n (see FIG. 1).

  The arithmetic circuit 36b subtracts the audio signal Y141 of the microphone M1 from the audio signal Y221 of the microphone M2, thereby generating an audio signal in which the audio component from the speaker SP is reduced over the frequency bands G1 to Gn. The audio signals Y141 and Y221 for the sound emitted from the SP have the same phase and amplitude at the frequency f1, and in particular, an audio signal in which the audio component from the speaker SP in the frequency band G1 is canceled is generated. Similarly, by subtracting the sound signal Y142 of the microphone M1 from the sound signal Y222 of the microphone M2, a sound signal in which the sound component from the speaker SP is canceled in the frequency band G2 is generated. By subtracting the audio signal Y14n of the microphone M1 from the audio signal Y22n, an audio signal in which the audio component from the speaker SP is canceled in the frequency band Gn is generated, and all the subtraction results in each frequency band G1 to Gn are added. To do. Furthermore, by dividing the addition result by the frequency band division number n and averaging, an audio signal Ya in which the audio component from the speaker SP is canceled for each frequency band G1 to Gn is generated.

  The averaging process can obtain substantially the same effect as the frequency division of the first embodiment without providing the bandpass filters 331 to 33n and 341 to 34n, and the averaging process will be described below. Detailed description.

  In the averaging process, the difference between the sound signals of the microphones M1 and M2 whose phases and amplitudes are matched for the frequency bands G1 to Gn with respect to the sound emitted from the speaker SP is separated into the frequency bands G1 to Gn. This is a process of simply adding without passing through the filter means, and dividing and averaging by dividing the frequency band by the division number n. The filter means is not required, and the circuit scale, cost and size can be reduced. . The following [Equation 5] is a mathematical expression for deriving the speaker sound cancellation amount Zmave by the averaging process.

Here, Zmk is the amount of cancellation of speaker sound by the above subtraction processing of the sound signals of the microphones M1 and M2 whose phases and amplitudes are matched in the frequency band Gk. Specifically, Zm1 is the phase in the frequency band G1. , The amount of speaker sound cancellation by the subtraction processing of the audio signals of the microphones M1 and M2 having the same amplitude, and Zm2 is the subtraction of the audio signals of the microphones M1 and M2 having the same phase and amplitude in the frequency band G2. Zmn is the amount of speaker sound canceled by the above subtraction processing of the sound signals of the microphones M1 and M2 whose phases and amplitudes are matched in the frequency band Gn.

  For example, as shown in FIG. 17, measured values Ys1 and Ys2 of the amount of cancellation of speaker sound when the amount of cancellation of speaker sound is adjusted to be optimal at frequencies of 1 KHz and 3 KHz, respectively, are added, and the addition result is expressed as a frequency. The theoretical value Yt21 of the cancellation amount by the averaging process of this embodiment is obtained by dividing by the number of divided bands n = 2.

Further, FIG. 18 is divided into a frequency band G1 of 2 KHz or less in which the amount of cancellation of the speaker sound is optimized at a frequency of 1 KHz and a frequency band G2 of 2 KHz or more in which the amount of cancellation of the speaker sound is optimized at a frequency of 3 KHz. The measured value Ys21 of the speaker sound cancellation amount when the averaging process is performed, and the approximate curve Ys22 of the actual measurement value Ys21 are shown, and there is no frequency band with an extremely high cancellation amount, but the cancellation is performed in a wide frequency band. In the first and second embodiments, the estimation circuit 322 to 32n estimates the audio signal of the microphone M2 for each frequency f2 to fn. The sound signal of the microphone M1 is estimated for each of the frequencies f2 to fn, and the estimated result is The same arithmetic processing as described above may be performed on the digital value of the audio signal of the microphone M2 (for example, estimation circuits 322 to 32n are provided in the branch path of the audio signal Y12 of the microphone M1).

2 is a circuit configuration diagram of a signal processing unit of the communication device according to Embodiment 1. FIG. It is side surface sectional drawing which shows a structure same as the above. It is a perspective view which shows a structure same as the above. It is a circuit diagram which shows the structure of an audio | voice processing part same as the above. It is side surface sectional drawing which shows the structure of a microphone substrate same as the above. It is side surface sectional drawing which shows the structure of a bare chip same as the above. It is the (a) simplified top view and (b) simplified circuit diagram which show the structure of a microphone substrate same as the above. It is a circuit diagram of an impedance conversion circuit same as the above. (A)-(f) It is a signal waveform diagram of the signal processing part same as the above. It is a figure which shows the outline | summary of the linear interpolation same as the above. (A) (b) It is a phasor vector figure which shows the outline | summary of the linear interpolation same as the above. (A) (b) It is a signal waveform diagram of a signal processing part same as the above. (A) (b) It is a signal waveform diagram of a signal processing part same as the above. (A) (b) It is a signal waveform diagram of a signal processing part same as the above. It is a figure which shows the cancellation amount by the signal processing part same as the above. It is a circuit block diagram of the signal processing part of the telephone apparatus of Embodiment 2. The frequency characteristic of the theoretical value of the cancellation amount by the averaging process is shown. The frequency characteristic of the actual measurement value of the cancellation amount by the averaging process is shown.

Explanation of symbols

A Caller SP Speaker M1, M2 Microphone 10e Signal processor 30, 31 A / D converter circuit 322-32n Estimator circuit 331-33n Bandpass filter 341-34n Bandpass filter 351-35n Attenuator circuit 36a Arithmetic circuit 37 Timing controller

Claims (5)

A speaker that outputs the transmitted audio information, a first microphone that collects audio and outputs an audio signal, and a distance from the speaker that is farther than the first microphone, collects audio A second microphone that outputs an audio signal; and a signal processing unit that processes and transmits each audio signal output by the first and second microphones;
The signal processing unit
First A / D conversion means for A / D converting an audio signal output from the first microphone;
The first A / D conversion means converts the sound signal from the first microphone using the transmission time of the sound wave of the first frequency corresponding to the difference in distance between the first and second microphones and the speaker as a delay time. Second A / D conversion means for A / D converting an audio signal from the second microphone when the delay time has elapsed from the timing of A / D conversion;
The phase difference of the second frequency component with respect to the first frequency component of the audio signal collected by one of the first and second microphones is used as a shift time, and the audio signal output from one microphone is a digital signal. An estimation means for estimating an audio signal output from one microphone at a timing shifted from the timing at which the A / D conversion means performs A / D conversion based on the output of the A / D conversion means for converting to ,
A process of matching the output levels of the first and second microphones with respect to the sound from the speaker for each frequency band including the first and second frequencies, respectively, in each of the one A / D conversion means and the estimation means Level adjusting means applied to at least one of the output and the output of the other A / D converting means;
And a computing unit that outputs a difference between the audio signals of the first and second microphones that have passed through the level adjusting unit.   The estimation means uses the phase difference of the second frequency component relative to the first frequency component of the audio signal collected by the second microphone as a shift time, and based on the output of the second A / D conversion means 2. The communication apparatus according to claim 1, wherein a voice signal output from the second microphone is estimated at a timing shifted by the shift time from a timing at which the second A / D conversion means performs A / D conversion.   The estimation means uses the phase difference of the second frequency component with respect to the first frequency component of the audio signal collected by the first microphone as a shift time, and based on the output of the first A / D conversion means 2. The communication apparatus according to claim 1, wherein a voice signal output from the first microphone is estimated at a timing shifted by the shift time from a timing at which the first A / D conversion means performs A / D conversion.   The estimation means approximates that the audio signal is moving along a straight line connecting two digital values continuously output from one A / D conversion means, and performs the estimation operation based on the straight line. The communication device according to any one of claims 1 to 3.   A first filter means for passing a frequency band including a first frequency from the output of the one A / D conversion means, and further allowing a frequency band including a second frequency to be passed from the output of the estimation means; And a second filter means for separating and passing the output of the A / D conversion means for each frequency band, and the computing means passes through the first and second filter means and the level adjusting means. 5. The call device according to claim 1, wherein a difference for each frequency band of the audio signal of the second microphone is added and output.

Images