JP2004120717A - Voice input system and communication system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a voice input system and a communication system capable of assuring clear voice input, even in a noisy surrounding. <P>SOLUTION: The voice input system comprises an echo canceller section 120 for subtracting a pseudo-echo signal generated by an adaptive filter 121, based on a voice signal input to a first input terminal, from a transmission voice signal input to a second input terminal, thereby removing echo components in the transmission voice signal; a main microphone 151 for receiving sounds mainly from an target sound source; a reference microphone 152, positioned at a predetermined distance to the main microphone 151 for mainly receiving sounds other than those from the target sound source; and an voice delay section 180 for giving predetermined delays to the transmission voice signal from the main microphone 151. The delay time in the voice delay section 180 is set to be within the delay time of the echo components which are removed by the echo canceller section 120. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a voice input system and a communication system, and more particularly to a voice input system and a communication system suitable for securing a clear voice input even in an environment where there is noise around the communication device.
[0002]
[Prior art]
2. Description of the Related Art As a voice input system used in a communication system, there is a technique called a noise canceller for performing clear voice input even in an environment where noise is present around a communication device.
The following techniques (1), (2) and (3) are widely known as techniques of this noise canceller.
(1) The first and second microphones are spaced apart from each other, and the difference between the delay signal of the output of the second microphone and the output of the first microphone is formed by first adding means. The difference between the one microphone output and the delay signal is formed by the second adding means. Patent Document 1 below describes that a noise component is canceled by calculating a difference between the two.
[0003]
(2) The generation of an audio signal from which a noise component has been removed from the input signal from the level difference between the audio signal and the noise signal included in the input signal from the first microphone is described in Patent Document 2 below. ing.
[0004]
(3) An audio signal is input from the first microphone, a noise / audio signal is input from the second microphone, a transfer function of noise included in the input signal from the second microphone is estimated, and a pseudo noise component is calculated. The generation of an audio signal from which a noise component has been removed by subtraction from an input signal from a first microphone is described in Patent Document 2 below.
[Patent Document 1]
JP-A-11-18186
[Patent Document 2]
JP-A-9-36663
[0005]
[Problems to be solved by the invention]
The above-mentioned noise canceling techniques (1) to (3) have the following problems (a), (a) and (c), respectively.
(A) According to the above technique (1), in an environment where noise is emitted from one specific direction on all four sides and only the target sound is emitted from another specific direction, the target sound has good S / N. You can get it.
However, the noise is emitted from an angle different from a specific direction, the noise is reflected from an obstacle and enters at an angle different from the specific direction, or is the same direction as the target sound. If the noise is caused by the background, there is a problem that the noise cannot be canceled.
[0006]
(A) According to the technique (2), when there is a considerable level difference between the collected voice signal and the noise signal, the noise component can be removed.
However, when there is almost no level difference between the audio signal and the noise signal, or when the noise signal is larger, a problem occurs that the noise component cannot be removed.
[0007]
(C) In the above technique (3), if the pseudo noise component estimated from the transfer function of the noise included in the signal collected from the second microphone matches the noise component collected from the first microphone, Can be canceled.
However, when the sound quality or volume of the noise suddenly changes, the estimated pseudo noise component does not match the noise component collected from the first microphone, and the audio signal after removing the pseudo noise component is There is a new problem that the sound becomes unnatural including different noises.
[0008]
The present invention has been made in view of the above-described problems, and regardless of the direction in which a target sound source and a noise source are generated, and irrespective of the level difference between the target sound and the noise, the sound volume and the sound quality are further improved. Even if the noise suddenly changes, it can remove these noises satisfactorily and realize a voice input system and communication system suitable for securing clear voice input even in an environment where there is noise around. The purpose is to do.
[0009]
[Means for Solving the Problems]
Therefore, the inventor of the present application has conducted intensive studies in order to solve the above-mentioned problems, and as a result, instead of using a system related to a known noise canceller, the echo canceller is operated as a noise canceller different from its intended use, so that We have found that we can solve the problem.
In addition, the present inventors have also studied the peripheral circuit configuration when operating the echo canceller as a noise canceller, and as a result, have reached the present invention.
[0010]
That is, the present invention as means for solving the above-mentioned problems has the following configuration.
(1) According to the first aspect of the present invention, a pseudo echo signal generated by an adaptive filter based on an audio signal input to a first input terminal is subtracted from a transmission audio signal input to a second input terminal. Thus, an echo canceller unit that removes an echo component mixed in the transmission audio signal, a main microphone that collects sound from a target sound source, and a target microphone that is arranged at a predetermined distance from the main microphone. A reference microphone for collecting sound mainly from a sound source other than a sound source; and a sound delay unit for giving a predetermined delay to a transmission sound signal from the main microphone, wherein the echo signal is transmitted from the reference microphone to the echo canceller. A first input terminal of the main unit, and a transmission audio signal from the main microphone delayed by the audio delay unit; Input to the second input terminal of the range of the delay time of the echo component the echo canceller unit is removed, adjust the delay time in the sound delay unit, a voice input system, characterized in that.
[0011]
In the invention having such a configuration, the main microphone collects sound mainly from a target sound source, the reference microphone collects sound mainly from a source other than the target sound source, and converts an audio signal from the reference microphone into an echo canceller unit. And a transmission audio signal from the main microphone delayed by the audio delay unit is input to a second input terminal of the echo canceller unit, and a delay of an echo component to be removed by the echo canceller unit. By adjusting the delay time in the audio delay unit within the time range, the pseudo echo signal generated by the adaptive filter based on the audio signal input to the first input terminal is transmitted to the second input terminal. By subtracting from the trusted audio signal, sound (noise) from other than the target sound source mixed in the transmission audio signal is treated and removed in the same manner as the echo component. It can be.
[0012]
As a result, regardless of the direction in which the target sound source and the noise source are generated, and regardless of the level difference between the target sound noise and the noise, even if the volume or the sound quality suddenly changes, these noises , And clear voice input can be ensured even in an environment where noise is present in the surroundings.
[0013]
(2) According to a second aspect of the present invention, a pseudo echo signal generated by an adaptive filter is subtracted from a transmission audio signal input to a second input terminal based on an audio signal input to a first input terminal. Thus, an echo canceller unit that removes an echo component mixed in the transmission audio signal, a main microphone that collects sound from a target sound source, and a target microphone that is arranged at a predetermined distance from the main microphone. A reference microphone that collects sound mainly from a sound source other than a sound source, an audio delay unit that applies a predetermined delay to an audio signal for transmission from the main microphone, and a reception signal from an external system, and the echo canceller unit. An interface for transmitting the processed transmission audio signal to an external system; and a reception signal obtained via the interface as audio. And an audio signal from the reference microphone is input to a first input terminal of the echo canceller unit, and a transmission audio signal from the main microphone delayed by the audio delay unit is transmitted to the echo canceller. A delay time in the voice delay unit that is input to a second input terminal of the unit and is within a range of a delay time of an echo component to be removed by the echo canceller unit.
[0014]
In the invention having such a configuration, the main microphone collects sound mainly from sound from the target sound source, the reference microphone collects sound mainly from sound other than the target sound source, and converts an audio signal from the reference microphone into an echo canceller unit. And a transmission audio signal from the main microphone delayed by the audio delay unit is input to a second input terminal of the echo canceller unit, and a delay of an echo component to be removed by the echo canceller unit. By adjusting the delay time in the audio delay unit within the time range, the pseudo echo signal generated by the adaptive filter based on the audio signal input to the first input terminal is transmitted to the second input terminal. By subtracting from the trust audio signal, the sound (noise) from other than the target sound source and the sound from the speaker (howling sound) mixed in the transmission audio signal are reduced. It can be removed handled similarly to the over component.
[0015]
As a result, regardless of the direction in which the target sound source and the noise source are generated, and regardless of the level difference between the target sound noise and the noise, even if the volume or the sound quality suddenly changes, these noises , And a clear voice input can be ensured even in an environment where there is ambient noise or sound from a speaker.
[0016]
(3) The invention according to claim 3, wherein the main microphone and the reference microphone are each configured to perform desired sound collection by a shielding member. An audio input system according to any one of the above.
[0017]
In the invention having such a configuration, in the audio input system according to the above (1) or (2), the main microphone collects sound mainly from the target sound source by the shielding member, and the reference microphone mainly forms the target sound source by the shielding member. It is configured to collect sound from other sources.
[0018]
As a result, in the above (1) or (2), regardless of the direction in which the target sound source and the noise source are generated, and regardless of the level difference between the target sound and noise, the volume and sound quality change suddenly. Noise, the noise can be removed satisfactorily, and a clear voice input can be ensured even in an environment where there is ambient noise or sound from a speaker.
[0019]
(4) The invention according to claim 4 is characterized in that the main microphone and the reference microphone are each configured to perform desired sound collection by directivity. An audio input system according to any one of the above.
[0020]
In the invention having such a configuration, in the voice input system according to the above (1) or (2), the main microphone picks up sound mainly from a target sound source by directivity, and the reference microphone picks up sounds other than the target sound source by directivity. The sound is mainly collected.
[0021]
As a result, in the above (1) or (2), regardless of the direction in which the target sound source and the noise source are generated, and regardless of the level difference between the target sound and noise, the volume and sound quality change suddenly. Noise, the noise can be removed satisfactorily, and a clear voice input can be ensured even in an environment where there is ambient noise or sound from a speaker.
[0022]
(5) The invention described in claim 5 is characterized in that the main microphone and the reference microphone are each provided with a shielding member, and furthermore, are configured to perform desired sound collection in combination with directivity. The voice input system according to any one of claims 1 and 2.
[0023]
According to the invention having such a configuration, in the voice input system according to the above (1) or (2), the main microphone collects sound mainly from a target sound source by a combination of a shielding member and directivity, and the reference microphone shields the sound. The combination of the member and the directivity is configured to mainly collect sound from a sound source other than the target sound source.
[0024]
As a result, in the above (1) or (2), regardless of the direction in which the target sound source and the noise source are generated, and regardless of the level difference between the target sound and noise, the volume and sound quality change suddenly. Noise, the noise can be removed satisfactorily, and a clear voice input can be ensured even in an environment where there is ambient noise or sound from a speaker.
[0025]
(6) The invention according to claim 6 is a communication system for performing wired or wireless communication, wherein the voice input of at least one side of the communication is performed by the voice input system according to any one of claims 1 to 5. A communication system characterized by using:
[0026]
In the invention having such a configuration, a configuration is employed in which a voice input system of any of (1) to (5) is used for voice input of at least one side of communication with respect to a communication system that performs wired or wireless communication. Have been.
[0027]
As a result, in a communication system that performs wired or wireless communication, regardless of the direction in which the target sound source and the noise source are generated, and regardless of the level difference between the target sound and noise, the volume and sound quality are suddenly reduced. Even for changing noise, these noises can be satisfactorily removed, and a clear voice input can be ensured even in an environment where there is ambient noise or sound from a speaker.
[0028]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, a voice input system and a communication system according to an embodiment of the present invention will be described in detail with reference to the drawings.
In the following embodiments, a communication system using wireless communication will be described as a specific example.
[0029]
FIG. 1 is a block diagram showing an electrical configuration of the first embodiment of the present invention. FIG. 2 is a block diagram showing an original method of using an echo canceller unit, which is a main component of the voice input system according to the first embodiment of the present invention.
[0030]
First, a system using an echo canceller as an original application will be described with reference to FIG.
Reference numeral 10 denotes a telephone interface, and a wired telephone line or a wireless telephone line is connected to the line side (right side in the figure), and the other end is connected to a microphone and a speaker via an echo canceller unit.
[0031]
Reference numeral 20 denotes an echo canceller unit that removes an echo component mixed in the transmission audio signal.
The echo canceller unit 20 transmits the pseudo echo signal generated by the adaptive filter 21 based on the audio signal input to the first input terminal (reception input terminal) 20a to the transmission signal input to the second input terminal (transmission input terminal) 20b. The subtracter 22 subtracts from the trusted audio signal, thereby removing the echo component mixed in the transmission audio signal from the microphone.
[0032]
Numeral 30 denotes an audio amplifier for amplifying the audio signal for reception, numeral 40 a speaker for converting the audio signal amplified by the audio amplifier 30 into an electric / acoustic signal and outputting it as a sound, and numeral 50 a sound from a target sound source (transmitting sound). This is a microphone provided for converting sound for transmission by a user into sound / electricity.
Note that the microphone 50 performs sound / electric conversion not only on the sound from the target sound source but also on the reception sound from the speaker 40 via various reflection paths.
[0033]
In such a configuration, the receiving sound from the line side passes through the echo canceller unit 20, is amplified by the audio amplifier 30, is then subjected to electric / acoustic conversion by the speaker 40, and is output as sound (receiving sound). .
When the speaker 40 and the microphone 50 are located in a certain proximity, the microphone 50 picks up sound (noise) from the speaker 40.
[0034]
Further, the microphone 50 generates sound for transmission by performing sound / electric conversion of the sound for reception output from the speaker 40 together with the sound (transmission sound) from the target sound source.
The transmission audio signal is a signal in which the echo component has been removed by removing the pseudo echo signal generated from the reception audio signal by the subtraction unit 22 of the echo canceller unit 20.
Then, the transmission audio signal from which the echo component has been removed is transmitted to the line side via the telephone interface 10.
[0035]
The above-described echo canceller unit 20 is used only for the purpose of removing the echo component when the receiving voice output from the speaker 40 is mixed into the microphone 50 via the reflection path to generate the echo component. Have been.
For this reason, the echo canceller unit 20 is configured so that the adaptive filter 21 can generate a pseudo echo signal within the delay time in consideration of the delay time of the echo component that goes from the speaker 40 to the microphone 50.
[0036]
Hereinafter, a voice input system and a communication system according to the present embodiment will be described with reference to FIG.
Reference numeral 101 denotes a control unit for controlling the whole, which sets a delay time of an audio delay unit described later, performs various levels of control and timing control for an echo canceller unit described later, and further controls a wireless interface. On the other hand, various levels of control and output control of alarm sound and the like are performed.
[0037]
110 is a wireless interface controlled by the control unit 101.
The wireless interface 110 has a wireless device connected to the right side (line side) in the figure, and the other end connected to a microphone via an echo canceller unit and a speaker via an audio amplifier.
The wireless interface 110 also corrects the frequency characteristics of the transmission signal and the reception signal.
[0038]
Reference numeral 120 denotes an echo canceller that removes an echo component mixed in the transmission audio signal, and has a circuit configuration equivalent to that of the echo canceler 20 shown in FIG. In the echo canceller section 120, a subtractor 122 subtracts the pseudo echo signal generated by the adaptive filter 121 from the transmission audio signal input to the second input terminal 120b based on the audio signal input to the first input terminal 120a. As a result, the echo component mixed in the audio signal for transmission from the microphone is removed.
[0039]
Note that a commercially available echo canceler LSI can be used as the echo canceller unit 120. However, unlike a normal use method (connection method), an audio signal from a reference microphone described later is supplied to the first input terminal. 120a.
[0040]
Reference numeral 130 denotes an audio amplifier that amplifies a reception sound obtained from the wireless device via the wireless interface 110, and reference numeral 140 denotes a speaker that performs electrical / acoustic conversion of the audio signal amplified by the audio amplifier 130 and outputs the converted audio signal.
[0041]
Reference numeral 151 denotes a main microphone provided for performing sound / electric conversion mainly on a sound from a target sound source (transmission sound by a transmitter) S (1).
The main microphone 151 is configured to collect sound mainly at a target sound source by using one or both of the directivity and the shielding member 161.
[0042]
Note that the microphone 151 includes not only the sound S (1) from the target sound source but also the surrounding noises N (1) to N (n) and the sound to be received from the speaker 140 via various reflection paths. Sound / electric conversion.
[0043]
152 is a reference microphone.
The reference microphone 152 is disposed at a predetermined interval from the main microphone 151, and is arranged so that sound from other than the target sound source (mainly, ambient noises N (1) to N (n)) is centered from the target sound source. The sound collection is performed in a state where the sound S (1) is excluded as much as possible.
The reference microphone 152 is configured to collect sound with a focus on ambient noise by using one or a combination of the directivity and the shielding member 162.
[0044]
Reference numeral 171 denotes a microphone amplifier that amplifies the transmission audio signal from the main microphone 151 to a predetermined level, and 172 denotes a microphone amplifier that amplifies the audio signal from the reference microphone 152 to a predetermined level.
Note that the audio signal from the reference microphone 152 amplified by the microphone amplifier 172 is input to the first input terminal 120a of the echo canceller unit 120.
[0045]
Reference numeral 180 denotes an audio delay unit for giving a predetermined delay to the transmission audio signal from the main microphone 151, and the transmission audio signal from the main microphone 151 delayed by the audio delay unit 180 is output to the echo canceller unit 120. It is input to the two input terminals 120b.
[0046]
In the above configuration, the control unit 101 controls the delay time of the audio delay unit 180 so as to match (include) the delay time of the echo component that can be removed by the echo canceller unit 120. I do.
[0047]
A wireless device 200 is a communication device that constitutes one of the communication systems. Here, the voice input system according to the present embodiment is connected via a wireless interface 110. Reference numeral 300 denotes a wireless device that constitutes the other side of the communication system. The voice input system according to the present embodiment is not connected, and a caller wirelessly communicates via the headset 310.
[0048]
Here, as a specific example, a speech input system and a communication system according to the present embodiment that are configured to accept an order through drive-through of a fast food restaurant will be described.
[0049]
In the above configuration, the main microphone 151 is installed so as to pick up the real voice of the customer who mainly comes through the drive-through (sound S (1) from the target sound source; hereinafter, referred to as “target sound”). .
In addition, the reference microphone 152 outputs the sound of the engine (N (1)) of the customer's vehicle coming to the drive-through, the sound of the cooling fan of the radiator (N (2)), or the sound of another vehicle traveling on the road. It is installed so as to collect ambient noise such as running noise (N (n)).
[0050]
In a drive-through provided with such a voice input system and a communication system, when a customer who comes to the drive-through places an order without getting out of the car, a driver brings the car close to the main microphone 151 in a so-called so-called manner. It is desirable to stop the car by closing the side.
[0051]
However, not all customers approach the main microphone 151 and stop. For this reason, there is a case where an interval of about 1 m or more is left between the customer in the driver's seat of the car and the main microphone 151.
[0052]
In such a case, the main microphone 151 not only outputs the target sound, but also the engine sound (N (1)) of the customer's automobile that came to drive through, the sound of the radiator cooling fan (N (2)), and furthermore, In addition, various ambient noises such as the running noise (N (n)) of other vehicles traveling on the road are also collected.
[0053]
Therefore, it is desirable to install the shielding member 161 so that various noises do not easily enter the main microphone 151.
Here, as the shape of the shielding member 161, a plate-shaped member is specifically mentioned, but it is preferable to appropriately select a shielding member 161 in consideration of the sound insulation effect, for example, a cylindrical member according to the use situation.
[0054]
On the other hand, the shielding member 162 is a line connecting the reference microphone 152 and the customer's mouth near the reference microphone 152 so that the actual voice (S (1)) of the customer who has come into the drive-through does not enter the reference microphone 152. It is desirable to install in such a manner as to intersect only with the real voice of the customer (S (1)).
As the shielding member 162, similarly to the shielding member 161, a specific shape such as a plate shape or a cylindrical shape may be mentioned, but it is desirable to appropriately select the shape in accordance with the use situation in consideration of the sound insulation effect. .
[0055]
Further, it is desirable that the distance between the reference microphone 152 and the main microphone 151 is 1.5 m at the longest.
From this, the maximum noise arrival time difference from when the noise first arrives at the main microphone 151 to when the same noise arrives at the reference microphone 152 is 1.5 / 340 ≒ 0 when the sound speed is 340 m / sec. .0044 seconds = 4.4 milliseconds.
[0056]
The echo canceller IC for the echo canceller unit 120 used in the experiment by the present inventor for this embodiment has a permissible time for canceling the echo in the range of 1 ms to 59 ms. .
[0057]
Therefore, in order to cancel the noise with the used echo canceller IC, 4.4 + 1 = 5.4 ms or more, and the target sound collected by the main microphone 151 in the range of 5.4 ms to 59 ms. + It is necessary to delay the noise signal.
[0058]
Therefore, it is necessary to use an audio delay unit 180 having an audio reproduction characteristic of low noise and low distortion rate and capable of setting an audio delay time of 5.4 ms or more.
In the experiment for the present embodiment, the experiment was performed by setting the audio delay time in the audio delay unit 180 to 20 milliseconds under the control of the control unit 101.
[0059]
That is, as the audio delay unit 180, a digital audio delay IC that delays the collected signal only for a certain set time and does not change the content of the signal as much as possible is preferable.
[0060]
Note that a reception signal of a mobile phone or the like is input to the first input terminal 120a of the echo canceller unit 120 in its intended use, but in the present embodiment, the sound is collected by the reference microphone 152 (N (1)). ) To (N (n)), the surrounding noise picked up by the reference microphone 152 is processed in the same manner as in the case of the received signal of the mobile phone.
[0061]
The second input terminal 120b of the echo canceller unit 120 has the real voice (S (1)) and the noise of (N (1)) to (N (n)) of the customer coming to the drive-through collected by the main microphone 151. Are input as audio signals (S (1)) ′ and (N (1)) ′ to (N (n)) ′ delayed by the delay time imposed by the audio delay unit 180.
[0062]
In the echo canceller unit 120, an audio signal (noise around (N (1)) to (N (n)) input to the first input terminal 120a is converted into a reference digital signal by an A / D converter (not shown). After conversion into an audio signal, the adaptive filter 121 retains and updates information.
[0063]
The transmission audio signals ((S (1)) 'and (N (1))' to (N (n)) 'input to the second input terminal 120b are also A / D converters (not shown). Is converted into a digital audio signal for transmission.
Then, the digital audio signal for transmission is compared with the digital audio signal for reference held in advance, and when similar information is mixed, it is determined that an echo is present, and the subtraction unit 122 is used to determine whether the echo signal has been received. , And cancels the digital audio signal for transmission.
Thereafter, the transmission digital audio signal is converted back to an analog signal by a D / A converter (not shown).
[0064]
By these series of operations, the target sounds (S (1)) ′ and (N (1)) ′ to (N (n)) ′ collected and delayed by the main microphone 151 are transmitted by the echo canceller unit 120. When output from the output terminal, noise cancellation is performed to the extent that only the target sound (S (1)) 'can be heard, and as a result, (N (1))' to (N (n)) ) 'Will be obtained.
[0065]
The target sound (S (1)) ′ signal is input to the wireless interface 110, and after being corrected for frequency characteristics or adjusted to the transmission input level of the wireless device 200 as necessary, is sent to the wireless device 200.
Here, the target sound (S (1)) ′ signal is added as a modulation to a transmission radio wave by the wireless device 200 and transmitted to the wireless device 300.
[0066]
This transmission radio wave is received by the wireless device 300, demodulated into an audio signal, and output as audio from the headphone section of the headset 310.
The headset 310 is worn by a store clerk equipped with a drive-through to listen to customer orders, but ambient noises (N (1)) to (N (n)) are canceled. As a result, only the voice of the customer (the target sound) is heard, and the exchange of the order and the like is performed simply and without fail, thereby increasing the work efficiency.
[0067]
The salesperson's conversation is picked up by the microphone section of the headset 310 and sent to the wireless device 300. The clerk uses the headset 310 to talk in a quiet store, and the microphone of the headset 310 is at his mouth.
Therefore, the conversation voice is a clear voice that is almost completely free of noise.
[0068]
Then, the conversation voice of the clerk is added as a modulation to the transmission radio wave by the wireless device 300 and transmitted. This transmission radio wave is received by the wireless device 200, demodulated into an audio signal, and sent to the wireless interface 110.
[0069]
The audio signal has its frequency characteristics corrected and its level adjusted by the wireless interface 110, and is sent to the audio amplifier 130.
Then, the audio signal is amplified by the audio amplifier 130 with a preset gain and output to the speaker 140.
Here, since the speaker 140 reproduces the voice of the clerk at an appropriate volume, the customer can hear the clear voice of the clerk.
[0070]
The output sound from the speaker 140 is also collected by the reference microphone 152 and the main microphone 151, but is canceled in the same manner as in the case of surrounding noise ((N (1)) to (N (n)). It will not be echoed or howling.
Further, in this case, the audio signal output from the speaker 140 does not pass through the inside of the echo canceller unit 120, which is different from the typical and original usage of the conventional echo canceller unit. Is a point.
[0071]
In performing the above operations, the control unit 101 sets the delay time of the audio delay unit 180, and performs various levels of control and timing control on the echo canceller unit 120 as necessary.
Further, the control unit 101 controls the wireless interface 110 at various levels, controls output of an alarm sound and the like, processes various sensor information, and the like.
[0072]
Although not shown, an A / D converter is provided on the input side of the audio delay unit 180, and a D / A converter is provided on the output side of the audio delay unit 180.
Then, although not shown, an AD converter is provided at the second input terminal 120b of the echo canceller unit 120, and a DA converter is provided at the output side.
Therefore, by integrally configuring the echo canceller unit 120 and the audio delay unit 180, it is possible to eliminate the intermediate DA converter and AD converter, and a cost merit is obtained.
[0073]
Further, as an application of the voice input system and the communication system of the present embodiment, in addition to the drive-through system, in addition to wireless communication between a crane operator and a crane wire-laying worker, when a jet airliner moves on the ground It can be used for wireless communication under a noisy work environment, such as wireless communication between a pilot in a passenger aircraft cockpit and a worker moving a passenger aircraft with a tractor.
[0074]
In these cases, the main microphone may be provided in a normal microphone portion of the headset, and the reference microphone may be provided in any other position (such as a headband portion).
Further, it can be used for a voice input device such as a telephone terminal, a video conference device, and a car navigation device.
[0075]
As described above, in the present embodiment, as a new noise solving method different from the conventional one, by adopting the echo canceller by a method different from the original, regardless of the generation direction of the target sound source and the noise source, Noise can be removed satisfactorily, and clear voice input can be ensured even in an environment where noise is present in the surroundings.
[0076]
Further, regardless of the level difference between the target sound and the noise, the noise can be satisfactorily removed, and a clear voice input can be ensured even in an environment where noise is present in the surroundings.
Furthermore, even if the sound volume or sound quality changes suddenly, the noise can be satisfactorily removed, and a clear voice input can be ensured even in an environment where there is noise around.
[0077]
【Example】
For the voice input system described above, an LSI “MSM7731-02” manufactured by Oki Electric Co., Ltd. developed for mobile phones was used as the echo canceller unit 120, and the above was described so as to differ from the original connection method of this LSI. A circuit was configured by the connection of the embodiment.
Further, an experiment was performed using a digital audio delay IC for the audio delay unit 180 and setting the audio delay time to 20 milliseconds.
[0078]
Here, an experiment was performed using the engine sound of a diesel vehicle generated in a frequency band substantially equal to the frequency band of a human voice as noise.
As a result, the sound after the noise cancellation was so reduced that the engine sound of the car could not be heard, and only the voice of the customer was heard more clearly than the headset of the clerk.
[0079]
Therefore, the voice from the customer side is clear and does not contain noise enough to be recognized, so that the order contents of the customer who is in a place with a remote noise are not overlooked or mistakenly heard. Thus, it is possible to provide a drive-through system that is speedy and has high rotational efficiency without causing the customer to feel unpleasant.
[0080]
On the other hand, a similar experiment was performed for a conventionally known noise cancellation. As a result, it was confirmed that the sound from the customer side contained the engine sound of the diesel car so that it could be heard and recognized, and it was confirmed that the real voice from the customer side was difficult to hear.
That is, a great and clear difference was recognized between the present invention and the conventional art.
[0081]
In addition to the combination of the main microphone 151 and the shielding member 161 and the combination of the reference microphone 152 and the shielding member 162, for example, the main microphone 151 selects a high-sensitivity microphone and the reference microphone 152 selects a normal-sensitivity microphone. As described above, by setting the sensitivity of the main microphone 151 and the sensitivity of the reference microphone 152 to be appropriately different from each other, it becomes possible to obtain a more clear voice input for noise.
[0082]
That is, for example, in a use environment, the level of the sound S (1) from the target sound source and the sound (noise) from other sources than the target sound source exist at the same level, and the echo canceller unit 120 is connected to the first input terminal. The echo canceller unit 120 has a characteristic of performing echo cancellation only for a signal exceeding a certain level in the signal level input to the input unit 120a. A case will be described in which the signal can be attenuated by 20 dB and a specific level is set to a level lower by 15 dB than the level of the sound S (1) from the target sound source.
[0083]
Here, in order to prevent the sound S (1) from the target sound source from being canceled as much as possible, the level of the sound S (1) from the target sound source collected by the reference microphone 152 is determined by the main microphone 151. By setting the level to be lower than the level of the sound S (1) from the sound source by 15 dB, the echo canceller unit 120 will not cancel the sound.
[0084]
Although a more specific embodiment has been implemented to determine the effectiveness of the present invention, the sound S (1) from the target sound source is shielded by the shielding member 162 so that the sound S (1) from the target sound source is canceled as much as possible. Although the reference microphone 152 does not collect sound, the reference microphone 152 must also collect sound (noise) from a source other than the target sound source. Even if the reference microphone 152 is separated from the target sound source, the reference microphone 152 does not. There is a limit to attenuating the sound S (1) from the target sound source with only the side shield member 162.
[0085]
Therefore, an attempt was made to solve a case where the level of the sound S (1) from the target sound source attenuated by the shielding member 162 on the reference microphone 152 side was only about 10 dB.
That is, when the level of the sound S (1) from the target sound source attenuated by the shielding member 162 on the reference microphone 152 side is only about 10 dB, the attenuation amount is reduced by 5 dB.
[0086]
Here, the reference microphone 152 uses a microphone having normal sound pickup sensitivity, and the main microphone 151 uses a microphone that is 6 dB higher than the sound pickup sensitivity of the reference microphone 152.
In theory, the level of the sound S (1) from the target sound source collected by the reference microphone 152 is set to be 10 + 6 = 16 dB lower than the level of the sound S (1) from the target sound source collected by the main microphone 151. It becomes possible.
On the other hand, a measured value using a level meter was able to obtain similar results, and the sense of hearing was good.
[0087]
This also raises the level of sound from the source other than the target sound source picked up by the main microphone 151 by 6 dB. However, the cancellation capability of the echo canceller IC used this time is about 20 dB, so that 20-6 = It is possible to cancel sounds from sources other than the target sound source of 14 dB.
That is, even in an environment where the level of the sound S (1) from the target sound source and the sound (noise) from other sources than the target sound source exist at the same level, the signal input to the transmission input of the wireless interface 110 does not Since only the sound (noise) from other than the target sound source is attenuated by 14 dB while keeping the level of the sound S (1), almost only the sound S (1) from the target sound source is obtained.
[0088]
As described above, it has been found that the difference between the sensitivity level of the main microphone 151 and the sensitivity level of the microphone 152 contributes to the main microphone 151 obtaining a clearer voice input for noise.
[0089]
【The invention's effect】
As described above in detail according to the embodiment, according to the present invention, regardless of the generation direction between the target sound source and the noise source, and regardless of the level difference between the target sound noise, Input system and communication system suitable for elimination of noise even when the noise or sound quality suddenly changes, and for ensuring clear voice input even in an environment where noise is present in the surroundings Can be realized.
[Brief description of the drawings]
FIG. 1 is a functional block diagram illustrating an electrical configuration of a voice input system according to an embodiment of the present invention.
FIG. 2 is a configuration explanatory diagram showing a configuration of a main part of the voice input system according to the embodiment of the present invention;
[Explanation of symbols]
101: control unit
110: Wireless interface
120: Echo canceller section
130: Audio Amplifier
140: Speaker
151: Primary microphone
152: Reference microphone
161, 162: shielding member
171,172: Microphone amplifier
180: audio delay unit

Claims (6)

By subtracting the pseudo echo signal generated by the adaptive filter based on the audio signal input to the first input terminal from the audio signal for transmission input to the second input terminal, the echo mixed in the audio signal for transmission An echo canceller section for removing components,
A main microphone that picks up sound from the target sound source,
A reference microphone that is arranged at a predetermined interval from the main microphone and collects sound mainly from sounds other than the target sound source;
An audio delay unit that applies a predetermined delay to the audio signal for transmission from the main microphone,
An audio signal from the reference microphone is input to a first input terminal of the echo canceller unit, and a transmission audio signal from the main microphone delayed by the audio delay unit is input to a second input terminal of the echo canceller unit. The delay time in the audio delay unit is adjusted within the range of the delay time of the echo component to be removed by the echo canceller unit.
A voice input system characterized by the following. By subtracting the pseudo echo signal generated by the adaptive filter based on the audio signal input to the first input terminal from the audio signal for transmission input to the second input terminal, the echo mixed in the audio signal for transmission An echo canceller section for removing components,
A main microphone that picks up sound from the target sound source,
A reference microphone that is arranged at a predetermined interval from the main microphone and collects sound mainly from sounds other than the target sound source;
An audio delay unit that applies a predetermined delay to the transmission audio signal from the main microphone;
An interface for receiving a reception signal from an external system and transmitting a transmission audio signal processed by the echo canceller unit to the external system,
And a speaker that outputs a received signal obtained through the interface as audio.
An audio signal from the reference microphone is input to a first input terminal of the echo canceller unit, and a transmission audio signal from the main microphone delayed by the audio delay unit is input to a second input terminal of the echo canceller unit. The delay time in the audio delay unit is adjusted within the range of the delay time of the echo component to be removed by the echo canceller unit.
A voice input system characterized by the following. The main microphone and the reference microphone are each configured to perform desired sound collection by a shielding member,
The voice input system according to claim 1, wherein: The main microphone and the reference microphone are each configured to perform desired sound collection by directivity,
The voice input system according to claim 1, wherein: The voice input device according to claim 1, wherein the main microphone and the reference microphone are configured to perform desired sound collection by a combination of a shielding member and a directivity, respectively. system. A communication system for performing wired or wireless communication, wherein the voice input system according to any one of claims 1 to 5 is used for voice input on at least one side of communication.

Images