JP2016111653A - Echo cancellation device, base device, communication system, and echo cancellation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an echo cancellation device, a base device, a communication system, and an echo cancellation method for reducing a transmission bit rate by a network type echo cancellation processing, and improving resistance to a signal defect.SOLUTION: A voice signal to be transmitted from an echo cancellation device on a network to a base device is stored in the echo cancellation device, and when the voice signal is transmitted from the base device, a signal paired with the voice signal is specified from the voice signals stored in the echo cancellation device, and echo cancellation processing is performed by using the specified voice signal to avoid deterioration in the echo cancellation processing generated due to a packet loss.SELECTED DRAWING: Figure 8

Description

  The present invention relates to an echo cancellation apparatus that performs network-type echo cancellation processing, a base apparatus connected to the echo cancellation apparatus, a communication system including these, and an echo cancellation method thereof.

  In voice conferences and video conferences over a network, processing for suppressing echo and noise (hereinafter referred to as echo cancellation processing or EC processing) is performed for a comfortable call. The echo cancellation process will be described later. The echo cancellation process is performed using both the voice signal sent from the other party and the voice signal input at the local site. Echo cancellation processing is a state called double talk, in which conversations are made simultaneously between the two locations. That is, when the voice of the sender is input to the microphone in addition to the sneak sound from the speaker, the echo cancellation processing is sufficient for signal processing. There is a problem that does not work. For this reason, when double talk occurs, the processing is not affected by taking measures such as stopping the echo cancellation processing.

  As an embodiment of the remote conference to which the echo cancellation processing is applied, a case (case A: FIG. 1) processed by a special device installed at the site to be implemented, and equipment on the network that relays each video / audio signal (Case B: FIG. 2) is assumed.

  In case A (FIG. 1), an echo cancellation process is performed by the EC processing unit in each site apparatus (10A, 10B) on the signal sent at each partner site and the signal collected at its own site.

  In Case B (FIG. 2), echo cancellation processing is performed by the EC processing unit 21 and the EC processing unit 22 of the echo cancellation device 20 on the network. That is, the base apparatus (10A, 10B) does not perform the echo cancellation process, but implements it with the echo cancellation apparatus 20 placed on the network. Therefore, in case B, the processes executed at each base are only simple processes such as AD conversion and transmission / reception processes.

FIG. 3 is a diagram for explaining detailed processing of case B. Each process is as follows.
[Treatment a]: The audio signal subjected to echo cancellation processing by the EC processing unit 21 is packetized and transmitted to the base device 10A at time (t).
[Process b]: The packet of process a arrives at the base apparatus 10A after (t + k) seconds and is received and processed.
[Processing c]: At time (t + k), the sound acquired by the base device 10A is packetized.
[Process d]: The audio signal received in process b and the audio signal acquired by the base apparatus 10A in c are linked with time information such as (t + k) seconds as a set of processes.
[Processing e]: The associated audio signal is transmitted from the transmission / reception processing unit 12 to the echo cancellation apparatus 20 on the network.
[Processing f]: The transmission / reception processing unit 22 of the echo canceling apparatus 20 transmits the audio signals of the processing b and c sent in the processing e to the EC processing unit 21 as a set.
[Processing g]: The EC processing unit 21 processes the audio signal sent from the processing f and transmits it to the opposite base (not shown).

Next, echo cancellation processing will be described.
The echo cancellation process is mainly performed by blocking a signal and converting it into a frequency domain for each frame. For adaptive filters during echo cancellation processing, specific processing in the frequency domain is described in Patent Documents 1 and 2. In this process, the frame X (i) of the received signal (= reproduced signal) and the frame Y (i) of the sound collection signal are input. The past several to ten and several frames are used as the reception signal. However, i is a frame number. FIG. 4 shows a case where the past four frames are used. At time t0, the adaptive filter is updated using X (i-3) to X (i), Y (i). At time t1, the adaptive filter is updated using X (i-2) to X (i + 1), Y (i + 1).

  The EC processing unit has two internal state variables and disposable variables for performing echo cancellation processing. The internal state variable is a variable depending on the past frame, and includes an adaptive filter coefficient and X (i). The disposable variable is a variable that does not depend on the past frame, and includes Y (i).

JP 2004-349796 A JP 2006-101437 A

  Two events that are the subject of the present invention will be described below.

  The first problem is that, in the processing of Case B, a voice signal sent from the network side and received at the site is sent back to the network side at the site side.

  The second problem is that when signal loss such as packet loss occurs in a transmission path such as a network, voice for performing echo cancellation processing is lost, and echo cancellation processing is not performed appropriately. Also for the above-described double talk, the corresponding processing when the double talk occurs operates on the premise that the sound is not interrupted by the respective input / output signals, and thus this detection processing also does not work.

  FIG. 5 is a diagram for explaining deterioration of echo cancellation processing when a frame including an audio signal is lost on the network.

  When a frame is lost at time t0, it is assumed that the audio signal included in the lost frame is zero-padded as a post-action response. Filling X (i) with zero is equivalent to setting the speaker volume to zero in an actual acoustic system. Even if the volume is set to 0, echoes resulting from past frames up to X (i-1) reach the microphone. Therefore, padding Y (i) with zero is equivalent to setting the microphone volume to zero. In an actual acoustic system, if the frame is lost due to packet loss or the like even though the volume of the speaker / microphone is not changed, the EC processing unit regards that the acoustic system has suddenly changed in adaptive filter signal processing. Try to follow. The effect of this frame loss will remain in the internal state variable.

  Next, a case where there is no problem with the current frame but a past frame is missing will be described. The update of the adaptive filter uses past frame information X (i-3) to X (i-1). If there is a defect in X (i-2) as shown in FIG. 5, if the adaptive filter is updated, the defect propagates to the coefficient of the adaptive filter and remains. As a result, the adaptive filter cannot be updated at time t0 and time t1.

  In an actual product, the number of frames of X (i) is 10 or more and is not short. Therefore, the length of the section in which the adaptive filter cannot be updated is much longer than the length of the section in which the frame is missing. Of the echo cancellation processing, processing related to the internal state variable is similarly affected by frame loss even in processing other than the adaptive filter. Therefore, once a deficit occurs, the impact is significant.

  In Case B, the echo cancellation process is concentrated on the network. Therefore, in order to appropriately execute the echo cancellation process, it is necessary to acquire the audio signal to each site and the audio signal from each site with no packet loss, that is, in a perfect state. However, if the audio signal sent to the base as shown in [a] in FIG. 6 is lost when passing through the network and the (t + k) signal cannot be acquired by the transmission / reception processing unit 12 as shown in [b], matching is performed. The voice is silent. In this case, the sound sent back to the echo cancellation apparatus 20 does not include the original sound as in [c]. As a result, a signal including the missing voice is passed to the EC processing unit 21, and an appropriate filter cannot be formed.

  In order to solve the first and second problems, the present invention provides an echo cancellation device, a base device, a communication system, which reduces the transmission bit rate by network-type echo cancellation processing and improves resistance to signal loss, It is another object of the present invention to provide an echo cancellation method.

  In order to achieve the above object, the present invention holds an audio signal transmitted from an echo canceling device on a network to a base device in the echo canceling device, and the voice signal is transmitted from the base device when the audio signal is transmitted. A signal paired with the signal is identified from the audio signal held in the echo cancellation device, and the echo cancellation process is performed using the identified audio signal, thereby avoiding the deterioration of the echo cancellation process caused by the packet loss.

Specifically, the echo cancellation device according to the present invention is:
The collected sound collected from the sound is transmitted as a transmission signal to the network and communicated with the base device that emits the reproduced sound based on the received signal received from the network, and the sound was emitted when collecting the sound. An echo cancellation apparatus that performs echo cancellation processing on a network to remove the echo from the collected voice including the reproduced voice as an echo,
A storage unit for storing a reception signal of a reproduction sound to be emitted from the base device;
When the transmission signal from the base device arrives, the reception signal corresponding to the reproduced sound reproduced when the collected sound of the transmission signal is collected by the base device is extracted from the storage unit, and the transmission is performed. An echo cancellation processing unit for performing echo cancellation processing using the signal and the received signal from the storage unit;
It is characterized by having.

Specifically, the base device according to the present invention is:
Communicating with an echo cancellation device on the network that performs echo cancellation processing on the network to remove the echo from the collected voice that contains the reproduced voice that was emitted as an echo when collecting the voice. A base device that transmits sound and sound as a transmission signal to a network and emits reproduced sound based on a reception signal received from the network,
An information generation unit that grasps a reception signal corresponding to a reproduction sound reproduced when the collected sound of the transmission signal is collected, and transmits related information between the transmission signal and the reception signal to the echo cancellation device; It is characterized by having.

Specifically, the communication system according to the present invention is:
The collected sound that has been collected is transmitted to the network as a transmission signal, and the base unit that emits the reproduced sound based on the received signal received from the network, and the reproduced sound that is emitted when the sound is collected are echoed. An echo canceling device for performing echo cancellation processing on the network to remove the echo from the collected voice included as, and a communication system comprising:
The echo canceling device is
A storage unit for storing a reception signal of a reproduction sound to be emitted from the base device;
When the transmission signal from the base device arrives, the reception signal corresponding to the reproduced sound reproduced when the collected sound of the transmission signal is collected by the base device is extracted from the storage unit, and the transmission is performed. An echo cancellation processing unit for performing echo cancellation processing using the signal and the received signal from the storage unit;
Have
The base device is
An information generation unit that grasps a reception signal corresponding to a reproduction sound reproduced when the collected sound of the transmission signal is collected, and transmits related information between the transmission signal and the reception signal to the echo cancellation device; It is characterized by having.

Specifically, the echo cancellation method according to the present invention is:
The collected sound that has been collected is transmitted to the network as a transmission signal, and the base unit that emits the reproduced sound based on the received signal received from the network, and the reproduced sound that is emitted when the sound is collected are echoed. An echo canceling method in a communication system comprising an echo canceling device that performs echo cancellation processing on a network to remove the echo from the collected voice included as
Store the reception signal of the reproduced sound to be emitted by the base device in the storage unit of the echo cancellation device,
The reception signal corresponding to the reproduction sound reproduced when the collected sound of the transmission signal is collected by the base device is grasped, and related information between the transmission signal and the reception signal is notified to the echo cancellation device. ,
When the transmission signal from the base device arrives at the echo cancellation device, it corresponds to the reproduced sound reproduced when the collected sound of the transmission signal is collected by the base device based on the related information A reception signal is extracted from the storage unit, and echo cancellation processing is performed using the transmission signal and the reception signal from the storage unit.

  Since this echo cancellation device holds the reception signal to be reproduced by the base device, there is no need to send back the reception signal from the base device, and even if the reception signal delivered to the base device is lost (packet loss) in the middle EC processing can be performed using the held received signal. Therefore, the present invention can provide an echo canceling device, a base device, a communication system, and an echo canceling method that reduce the transmission bit rate by the network type echo canceling process and improve the resistance to signal loss.

  The echo cancellation apparatus according to the present invention further includes an error correction function that restores the loss of the transmission signal being transmitted using the parity added to the transmission signal by the base apparatus.

  The base apparatus according to the present invention further includes an error correction function that restores a loss of the received signal during transmission using the parity added to the received signal by the echo canceling apparatus.

  The communication system according to the present invention adds a parity to the transmission signal and the reception signal between the base device and the echo cancellation device, and restores the loss of the transmission signal and the reception signal being transmitted. It further has a function.

  The echo cancellation method according to the present invention adds parity to the transmission signal and the reception signal between the base device and the echo cancellation device, and restores the loss of the transmission signal and the reception signal being transmitted. It is characterized by.

  Since the echo cancellation apparatus and the base apparatus are equipped with an error correction (FEC) function, a signal can be restored even when a packet loss occurs. For this reason, it is possible to avoid a state in which the “acoustic system has suddenly changed” due to packet loss, and appropriate echo cancellation processing can be performed.

  The present invention can provide an echo canceling device, a base device, a communication system, and an echo canceling method that reduce the transmission bit rate by network-type echo canceling processing and improve resistance to signal loss.

It is a figure explaining the form of the communication system to which the echo cancellation process is applied. It is a figure explaining the form of the communication system to which the echo cancellation process is applied. It is a figure explaining an echo cancellation process. It is a figure explaining an echo cancellation process. It is a figure explaining the malfunction of an echo cancellation process. It is a figure explaining the malfunction of an echo cancellation process. It is a figure explaining the communication system which concerns on this invention. It is a figure explaining the communication system which concerns on this invention. It is a figure explaining the communication system which concerns on this invention. It is a figure explaining the communication system which concerns on this invention. It is a figure explaining the echo cancellation method which concerns on this invention. It is a figure explaining the echo cancellation method which concerns on this invention. It is a figure explaining the echo cancellation method which concerns on this invention. It is a figure explaining the echo cancellation method which concerns on this invention.

  Embodiments of the present invention will be described with reference to the accompanying drawings. The embodiments described below are examples of the present invention, and the present invention is not limited to the following embodiments. In the present specification and drawings, the same reference numerals denote the same components.

(Embodiment 1)
The present embodiment is a method in which an error correction (FEC) function is mounted on the network side and the base side, so that even when signal loss occurs, the signal is restored and appropriate echo cancellation processing can be performed.

  FIG. 7 is a diagram illustrating the communication system 301 of the present embodiment. The communication system 301 transmits the collected sound, which is collected sound, to the network as a transmission signal, and emits the reproduced sound based on the received signal received from the network, and the sound emission when collecting the sound. And an echo canceling device 20 for performing echo cancellation processing on the network for removing the echo from the collected voice including the reproduced voice as an echo. In FIG. 7, the base device 10B, which is the delivery destination of the collected sound collected by the base device 10A and is the source of the reproduction signal to be emitted by the base device 10A, is not shown.

  The communication system 301 has an error correction function that adds parity to the transmission signal and the reception signal between the base device 10A and the echo cancellation device 20 and restores the loss of the transmission signal and the reception signal that are being transmitted. . Specifically, as shown in FIG. 7, the base apparatus 10 </ b> A includes an FEC processing unit 15, and the echo cancellation apparatus 20 includes an FEC processing unit 25.

  The FEC processing unit 25 adds decoding information called parity to the signal when the echo cancellation apparatus 20 sends a signal (the received signal) addressed to the base apparatus 10A to the network. In the base device 10A, when there is a loss of a signal (the received signal) sent from the network side, the FEC processing unit 15 restores the lost signal using the parity information. Similarly, the FEC processing unit 15 also adds parity information to the signal transmitted from the base apparatus 10A (the transmission signal), so that the echo cancellation apparatus 20 also restores the signal transmitted from the base apparatus 10A. Is possible. As a result, since the echo cancellation apparatus 20 can acquire an audio signal that should be received, an appropriate echo cancellation process can be performed.

(Embodiment 2)
This embodiment is a method of detecting which signal is lost on the base side and complementing on the network side during EC processing. About the audio | voice signal transmitted from the network side, the information which part audio | voice was missing in the base side is created as metadata, and is transmitted to the network side. On the network side, the voice transmitted to the base side is retained. When the metadata sent from the base includes information on voice loss due to packet loss, the voice is restored by acquiring the voice signal from the buffer.

  FIG. 8 is a diagram illustrating the communication system 302 according to the present embodiment. The communication system 302 includes a base device 10A, an echo cancellation device 20, and a base device 10B (not shown) as in the communication system 301 of FIG.

The echo cancellation device 20
A storage unit 27 for storing a reception signal of a reproduced sound to be emitted from the base device 10A;
When the transmission signal from the base device 10A arrives, the base device 10A takes out the received signal corresponding to the reproduced sound reproduced when the collected sound of the transmission signal is collected from the base device 10A from the storage unit 27, and transmits the transmission signal. An echo cancellation processing unit 21 that performs echo cancellation processing using the signal and a received signal from the storage unit 27;
Have
The base device 10A
An information generation unit that grasps a reception signal corresponding to a reproduction sound reproduced when the collected sound of the transmission signal is collected, and transmits related information MD between the transmission signal and the reception signal to the echo cancellation apparatus 20 Have
In the present embodiment, it is assumed that the transmission / reception processing unit 12 includes the information generation unit.

  The EC process of this embodiment is assumed to be case C. In Case B, the base device 10A performed processing to send back the audio signal (the received signal) received from the network side to the echo cancellation device 20. On the other hand, in case C, the audio signal (the received signal) received by the echo canceling device 20 from the network side is stored in the storage unit 27 (process [a]), and the base device 10A receives the received audio signal (the above-mentioned signal). The process of returning the received signal) to the echo cancel device 20 is not performed. Then, the base device 10A performs the matching operation of the frames that are paired with the EC process, similarly to the case B, on the audio signal from the network and the audio signal collected by itself.

  In the case C, the base device 10A matches the matching information MD (the related information) associated with the combination of EC processing by, for example, the number and time assigned to the unit of the audio signal such as the packet number and the frame number. (Process [b] [c]). The base device 10A transmits the matching information MD as metadata to the echo cancellation device 20 (processing [d]).

  The echo cancellation apparatus 20 extracts the audio signal (the received signal) stored in the storage unit 27 based on the packet number described in the received matching information MD (process [e]). The echo canceling device 20 sets the audio signal (the transmission signal) transmitted from the base device 10A based on the matching information and the audio signal (the reception signal) extracted from the storage unit 27 as a set. 21 (process [f]). The EC processing unit 21 performs EC processing based on the transmission signal and the reception signal.

  Here, a case where a signal loss actually occurs will be described with reference to FIG. In this case, the received signal “a123” to be reproduced by the base device 10A is lost and is not reproduced by the speaker. For this reason, the sound collected by the base device 10A is a sound that "a123" does not wrap around. That is, the collected transmission signal “b456” is a sound that does not include a wraparound sound. The base apparatus 10A grasps which frame or packet the received signal is lost, and adds this information to the matching information MD (processing [b] [c]). The echo canceling device 20 refers to the signal loss information described in the matching information MD, and if there is a loss, performs signal complement processing on the audio signal collected by the base device 10A, and wraps around The audio signal includes audio (process [e ′]). Specifically, the echo cancellation apparatus 20 extracts the audio signal “a123” (the received signal) corresponding to the missing signal from the matching information MD from the storage unit 27, and superimposes the audio signal “b456” on the collected audio signal “b456”. Is an audio signal.

  The echo cancellation apparatus 20 simultaneously extracts an audio signal (the received signal) corresponding to the EC process from the matching information MD from the storage unit 27 (process [e]). Then, the echo canceling apparatus 20 sets the audio signal that has been subjected to the complementary processing and the audio signal extracted from the storage unit 27 as a set and passes the set to the EC processing unit 21. Thereby, the EC processing unit 21 can perform an appropriate filter calculation.

  In Case C, echo cancellation processing can be performed even if voice is lost along the route by sending metadata called matching information to the network side without returning the voice signal sent from the base as it is by the above processing. . As a result, echo cancellation processing can be realized without using the network bandwidth.

  In the above description, the echo cancellation device 20 refers to the signal loss information described in the matching information MD, and if there is a loss, the echo cancellation device 20 outputs a signal to the sound signal collected by the base device 10A. Complementary processing was performed to obtain an audio signal including wraparound audio (processing [e ′]). Unlike the above, the echo cancellation apparatus 20 does not process the voice on the base side, refers to the information of the signal loss described in the matching information MD, and if there is a loss, from the storage unit 27 With regard to the audio to be extracted, 0 signal may be embedded in the audio signal portion of the a123 portion, and the audio signal that does not include the wraparound audio transmitted from the base side may be set and passed to the EC processing unit 21. .

(Embodiment 3)
The FEC function described in FIG. 7 may be installed in the communication system of FIG. For example, the signal is normally restored by the FEC process, but when the signal cannot be restored by the FEC process, the signal is acquired from the storage unit of the echo canceling apparatus as illustrated in FIG. EC processing can be continued.

(Embodiment 4)
This embodiment is an operation in the case where both the reproduced sound and the collected sound are lost as shown in FIG.
In the second and third embodiments, a description will be given of a case where, when a voice played back at the base device 10A is lost due to a packet loss on a transmission path, the voice can be restored and the influence on EC processing can be minimized. did. In the present embodiment, a description will be given of a case where packet loss occurs in a complex manner, such as loss of a signal in both the forward path and the return path, and cannot be relieved by FEC, or metadata cannot be referred to.

  When the echo canceling apparatus 20 recognizes that both the reproduced voice and the collected voice are missing, the EC canceling apparatus 20 stops the EC processing for the signal section (time). By stopping the filter operation in this section and not updating the filter coefficient, the filter operation in the subsequent EC processing is not affected.

(Embodiment 5)
In the present embodiment, an echo cancellation method in the communication system described in the first to fourth embodiments will be described.

  In the echo cancellation method of the present embodiment, which signal is lost is described in matching information for performing echo cancellation processing, and the lost signal is restored from the signal buffered in advance on the network side based on the information. I am going to do that. For example, the lost packet number is described in the metadata as the packet loss information. The metadata is created at each site, and has information about matching between the audio acquired at the site and a signal that is a pair of EC processing, that is, a signal sent from the network. For example, the packet number of the signal acquired and packetized at the base side is a-10, and is sent from the network side including the voice signal included in the No. 10 packet and the voice to be paired with EC processing. If the signal is b-20, information that a-10 and b-20 correspond is written in the metadata.

  The metadata includes information on whether or not the signal transmitted from the network has lost a packet. Note that the information that forms a pair of metadata described in the metadata is not limited to association by packet number, but may be associated by reception time or the like.

  The created metadata is sequentially sent from the base side to the network side. The network side can grasp which packet is lost among the packets sent to the base side and extract the signal paired with the voice signal sent from the base side from the signal buffered at the time of transmission. It becomes possible.

  The echo cancellation method of the present embodiment has a function of introducing a FEC function on each of the network side and the base side to restore a missing signal in order to properly operate the echo cancellation process in the echo cancellation process on the network. May be provided. By introducing FEC in this method, tolerance to disappearance is increased, and the influence on EC processing can be further reduced.

  In the echo cancellation method of the present embodiment, when the signal cannot be restored even if FEC is introduced, the EC processing is stopped for the data portion including the lost signal, and the influence of the data loss is reduced by not updating the filter. To do.

  11 to 14 are diagrams for explaining the echo cancellation method of the present embodiment. The echo cancellation method of this embodiment is EC processing on the IP network in the configuration of case C having the FEC function.

FIG. 11 is a diagram for explaining the operation of the echo canceling apparatus when transmitting the reproduced sound from the network side to the base apparatus.
The reproduction signal is processed by the echo cancellation processing unit (EC unit) 101 of the network echo cancellation apparatus 100 and sent to the FEC transmission processing unit 102. The FEC transmission processing unit 102 performs FEC processing and calculates parity information. After the parity information calculation is performed, the reproduction signal and the parity are sent to the packet transmission processing unit 103.

  The packet transmission processing unit 103 packetizes the audio signal and parity. At this time, the packet transmission processing unit 103 divides the input signal into a certain size, and gives the packet number and time information acquired from the FEC transmission processing unit 102 for each divided data. Further, the packet transmission processing unit 103 creates table information that describes the combination of the packet number and the acquisition time.

  The packetized signal and table information are stored in the buffer unit 104. Further, the packetized signal is transferred to the transmission unit 105, and is transmitted from the transmission unit 105 to the base device.

FIG. 12 is a diagram for explaining the operation of the site device when transmitting playback audio from the network side to the site device.
The receiving unit 201 of the site processing apparatus 200 receives a signal transmitted from the network side. The packet reception processing unit 202 extracts an audio signal from the packet sent from the reception unit 201 and buffers it for a certain amount, and then aligns the audio signal based on the packet number and passes the audio signal to the FEC reception processing unit 203. Also, the packet reception processing unit 202 creates reception signal table information in which the time when the signals are aligned and the respective packet numbers are set, and passes them to the FEC reception processing unit 203.

  The FEC reception processing unit 203 restores the signal using the parity when the packet loss received from the received signal table information is found. When the restoration is successful, the FEC reception processing unit 203 inserts the restoration signal into the missing part and inserts the packet number corresponding to the table information. When the restoration fails, the FEC reception processing unit 203 generates a zero-padded signal (silence signal) in the signal, and adds the failed packet number and information indicating the failure to the received signal table. The added reception signal table is transferred to the meta information creation unit 209.

  A signal is passed from the FEC reception processing unit 203 to the DA processing unit 204. The DA processing unit 204 converts a digital signal into an analog format, and passes the signal to the output unit 205 as sound that can be output from a speaker. The output unit 205 outputs the transmitted signal from the output terminal.

FIG. 13 is a diagram for explaining the operation of the base device when the collected sound is transmitted from the base device to the network side.
The input unit 206 receives the collected sound from a microphone or the like, and passes the collected sound to the AD conversion unit 207. The AD conversion unit 207 converts the received audio signal from analog to digital, and passes the signal to the FEC transmission processing unit 208. The FEC transmission processing unit 208 receives the transmitted signal and calculates parity information. The parity information is sent to the meta information creation unit 209 together with the audio signal.

  The meta information creation unit 209 creates transmission signal table information. Specifically, the time received from the FEC transmission processing unit 208 and the packet number assigned by the packet transmission processing unit 210 are assigned to each audio signal divided at regular intervals by the packet transmission processing unit 210, and the transmission signal table information is Created. Here, the meta information creation unit 209 refers to the created transmission signal table and the reception signal table information sent from the FEC reception processing unit 203, and each packet that is a combination of EC processing from each time information. A combination of numbers is extracted and a matching table is created. The matching table also includes packet loss information described in the received signal table. Note that the time information described in each table that defines a combination of EC processes is always uniquely determined because each process in the base processing apparatus 200 is processed with the same clock. The created matching table is also passed to the packet processing unit 210.

  The packet processing unit 210 receives an audio signal from the meta information creation unit 209, divides the received signal at regular intervals, and assigns a packet number to each. Further, the parity created by the FEC transmission processing unit 208 is made redundant to the packet here. The packetized signal is passed to the transmission unit 211. The matching table is also packetized by the packet transmission processing unit 210 and sent to the transmission unit 211. The transmission unit 211 transmits the packet transmitted from the packet transmission processing unit 210 and the matching table to the echo cancellation apparatus 100.

FIG. 14 is a diagram for explaining the operation of the echo canceling device when the collected sound is transmitted from the base device to the network side.
The reception unit 106 of the network echo cancellation apparatus 100 receives the packet transmitted from the site processing apparatus 200 and transmits the packet to the packet reception processing unit 107. The packet reception processing unit 107 extracts a voice signal, a parity, and a matching table from the packet after buffering the packet by a certain amount. The packet reception processing unit 107 arranges the audio signals in the order of packet numbers and passes the audio signal, parity, and reception packet number to the FEC reception processing unit 108. The packet reception processing unit 107 passes the matching table to the matching processing unit 109.

  The FEC reception processing unit 108 confirms the received packet number, and restores the missing signal using the parity from the audio signal and the parity information when the missing is found. When the restoration is successful, the FEC reception processing unit 108 inserts the restoration signal into the missing part and inserts the packet number corresponding to the table information. When the restoration fails, the FEC reception processing unit 108 inserts a zero padded signal (silence signal) into the signal. The failed packet number is transmitted to the EC processing unit 101. The FEC reception processing unit 108 sends the restored audio signal to the matching processing unit 109.

  Based on the sent matching table, the matching processing unit 109 extracts a signal that is a set of EC processing from the buffer unit 104 for each signal sent from the base device 200, and outputs a signal for EC processing. The set is sent to the echo cancel processing unit 101 as a set. In addition, the matching processing unit 109 extracts a corresponding audio signal from the buffer unit 104 for the packet number having the missing information in the matching table, and sends it to the matching processing unit 109 and also to the filter processing unit 110. The filter processing unit 110 synthesizes speech using the signal extracted from the buffer unit 104 and the audio signal (including the audio signal restored by the FEC reception processing unit) sent from the base device 200. Specifically, the missing sound stored in the buffer unit 104 is synthesized with a signal of only the collected sound that is not reproduced by the base device 200 due to a signal loss, and the base device 200 wraps and collects the sound. A signal equivalent to the case is synthesized.

  The EC processing unit 101 performs EC processing on the transmitted audio signal. The EC processing unit 101 determines when the data is incomplete based on the FEC processing failure information received from the FEC reception processing unit 108. If the data is incomplete, the EC processing unit 101 does not update the internal state. The internal state is a filter coefficient of the adaptive filter or a signal power estimation value that is sequentially estimated. The audio signal that has been subjected to the EC process is sent to the destination base device.

  The buffer unit 104 discards the audio signal held after a certain time after the EC processing ends in the EC processing unit 101.

(Effect of the invention)
According to the method of the present invention, voice loss is eliminated and a network use band for transmitting a signal can be kept low. In addition, it is possible to largely avoid the stop of the signal processing related to the update of the internal state variable inside the filter generated by the EC processing. Therefore, the performance degradation of EC processing due to frame loss can be greatly improved. As a result, in the network echo canceling process, even if a large number of losses occur in the transmission, the frequency of the influence can be greatly reduced, and the voice can be exchanged comfortably.

  According to the method of the present invention, it is possible to exchange an audio signal on which echo cancellation processing has been performed even with a general-purpose device without installing a special device equipped with an echo cancellation processing function on the site side.

10, 10A, 10B: Base device 20: Echo cancel device 12: Transmission / reception processing unit 15: FEC processing unit 21: EC processing unit 22: Transmission / reception processing unit 25: FEC processing unit 27: Buffer unit (storage unit)
100: Echo cancel device 101: EC processing unit 102: FEC transmission processing unit 103: Packet transmission processing unit 104: Buffer unit 105: Transmission unit 106: Reception unit 107: Packet reception processing unit 108: FEC reception processing unit 109: Matching process Unit 110: filter processing unit 200: site processing device 201: reception unit 202: packet reception processing unit 203: FEC reception processing unit 204: DA processing unit 205: output unit 206: input unit 207: AD conversion unit 208: FEC transmission processing Unit 209: meta information creation unit 210: packet transmission processing unit 211: transmission unit 301, 302: communication system

Claims (8)

The collected sound collected from the sound is transmitted as a transmission signal to the network and communicated with the base device that emits the reproduced sound based on the received signal received from the network, and the sound was emitted when collecting the sound. An echo cancellation apparatus that performs echo cancellation processing on a network to remove the echo from the collected voice including the reproduced voice as an echo,
A storage unit for storing a reception signal of a reproduction sound to be emitted from the base device;
When the transmission signal from the base device arrives, the reception signal corresponding to the reproduced sound reproduced when the collected sound of the transmission signal is collected by the base device is extracted from the storage unit, and the transmission is performed. An echo cancellation processing unit for performing echo cancellation processing using the signal and the received signal from the storage unit;
An echo canceling device comprising: The echo cancellation apparatus according to claim 1, further comprising an error correction function that restores a loss of the transmission signal being transmitted using the parity added to the transmission signal in the base apparatus.
Communicating with an echo cancellation device on the network that performs echo cancellation processing on the network to remove the echo from the collected voice that contains the reproduced voice that was emitted as an echo when collecting the voice. A base device that transmits sound and sound as a transmission signal to a network and emits reproduced sound based on a reception signal received from the network,
An information generation unit that grasps a reception signal corresponding to a reproduction sound reproduced when the collected sound of the transmission signal is collected, and transmits related information between the transmission signal and the reception signal to the echo cancellation device; A base device characterized by having. The base apparatus according to claim 3, further comprising an error correction function that restores the loss of the received signal being transmitted using the parity added to the received signal by the echo canceling apparatus.
The collected sound that has been collected is transmitted to the network as a transmission signal, and the base unit that emits the reproduced sound based on the received signal received from the network, and the reproduced sound that is emitted when the sound is collected are echoed. An echo canceling device for performing echo cancellation processing on the network to remove the echo from the collected voice included as, and a communication system comprising:
The echo canceling device is
A storage unit for storing a reception signal of a reproduction sound to be emitted from the base device;
When the transmission signal from the base device arrives, the reception signal corresponding to the reproduced sound reproduced when the collected sound of the transmission signal is collected by the base device is extracted from the storage unit, and the transmission is performed. An echo cancellation processing unit for performing echo cancellation processing using the signal and the received signal from the storage unit;
Have
The base device is
An information generation unit that grasps a reception signal corresponding to a reproduction sound reproduced when the collected sound of the transmission signal is collected, and transmits related information between the transmission signal and the reception signal to the echo cancellation device; A communication system comprising: It further has an error correction function of adding parity to the transmission signal and the reception signal between the base device and the echo cancellation device, and restoring a loss of the transmission signal and the reception signal being transmitted. The communication system according to claim 5.
The collected sound that has been collected is transmitted to the network as a transmission signal, and the base unit that emits the reproduced sound based on the received signal received from the network, and the reproduced sound that is emitted when the sound is collected are echoed. An echo canceling method in a communication system comprising an echo canceling device that performs echo cancellation processing on a network to remove the echo from the collected voice included as
Store the reception signal of the reproduced sound to be emitted by the base device in the storage unit of the echo cancellation device,
The reception signal corresponding to the reproduction sound reproduced when the collected sound of the transmission signal is collected by the base device is grasped, and related information between the transmission signal and the reception signal is notified to the echo cancellation device. ,
When the transmission signal from the base device arrives at the echo cancellation device, it corresponds to the reproduced sound reproduced when the collected sound of the transmission signal is collected by the base device based on the related information An echo cancellation method comprising: extracting a received signal from the storage unit; and performing echo cancellation processing using the transmission signal and the received signal from the storage unit.   The parity of the transmission signal and the reception signal between the base device and the echo cancellation device is added to restore the loss of the transmission signal and the reception signal being transmitted. Echo cancellation method.

Images