WO2007040222A1

WO2007040222A1 - Clock synchronization system and method in audio transmission system

Info

Publication number: WO2007040222A1
Application number: PCT/JP2006/319783
Authority: WO
Inventors: Akihiro Suzuki
Original assignee: Matsushita Electric Industrial Co., Ltd.
Priority date: 2005-10-05
Filing date: 2006-10-03
Publication date: 2007-04-12
Also published as: JP2007104347A

Abstract

In a clock synchronization system, a master device (11) corresponds to a main device and a sale device (12) corresponds to a remote device. The master device (11) transmits a synchronization packet in which time stamp information is embedded to a plurality of slave devices (12) via a network (13). Each of the slave devices (12) receives the synchronization packet, calculates a difference between the accumulated time stamp value To of the master device (11) and the accumulated time stamp value Ti of the slave device (12) itself, and adjusts frequency of a frequency-variable clock source (20) according to the difference (To - Ti). Furthermore, the slave device (12) controls frequency by using the fluctuation coefficient c proportional to the inclination degree of the difference (To - Ti). This eliminates expansion/shrinkage of the delay time of the audio in the audio transmission system and minimizes the audio output time difference between the reception devices.

Description

Specification

Clock synchronization system and method in audio transmission system

Technical field

TECHNICAL FIELD [0001] The present invention relates to a clock synchronization system and method provided in an audio transmission system composed of a main device and a remote device connected via a network, and in particular, a preset one main device (clock master) The present invention relates to a clock synchronization system and method for synchronizing each remote device (slave device) force S clock with reference to a synchronization packet transmitted separately from the voice packet from the device.

Background art

[0002] As the network environment is improved, communication services using the Internet Protocol (IP) network are expanding. As an example of utilizing these, voice transmission systems that transmit voice over an IP network are attracting attention. These systems are mainly used for broadcasting to remote locations and multipoint broadcasting such as multi-store commercial facilities.

[0003] In an audio transmission system composed of a plurality of devices connected via a network, audio data is packetized and transmitted. Therefore, it is necessary for the receiving apparatus to perform buffering until audio data that can be reliably output is available. At this time, the clock frequencies used by the transmission device and the reception device are the same, V, and the difference is that the audio buffer of the reception device overflows or underflows, and the output audio is interrupted. There was a problem that would end up.

[0004] To solve this problem, clock synchronization must be performed between the transmission device and the reception device. This method will be described with reference to FIG.

FIG. 12 is a diagram showing an example of a clock synchronization method in a conventional audio transmission system.

As shown in FIG. 12, a conventional audio transmission system 50 includes a master device 51 and a slave device 52, which are connected via a network 53. As the master device 51, one device is set in advance in the system. The master device 51 has a function of transmitting a packet containing its own clock signal information to another slave device 52 via the network 53. On the other hand, one or more slave devices 52 exist in the system, and A function of receiving the packet via the network 53;

[0006] The master device 51 includes a clock source 54, a time stamp extraction unit 55, and a packet transmission unit 56. If the clock synchronization is to be started, the time stamp extraction unit 55 transmits the time stamp information obtained by counting the clock frequency from the clock source 54 to the packet transmission unit 56. The packet transmission unit 56 packetizes the received time stamp information and transmits it to the network 52. There are various forms of communication protocols to be sent over the network, but here we consider an example using the IP protocol. In the case of a voice transmission system composed of a large number of slave devices 52, a plurality of slave devices can easily receive packets by transmitting packets from the master device 51 by IP multicast.

The slave device 52 includes a packet reception unit 57, a packet transmission interval calculation unit 58, a frequency variable clock source 59, a time stamp extraction unit 60, a packet reception interval calculation unit 61, and an operation correction unit 62. When the packet reception unit 57 receives a packet from the master device 51, the packet transmission interval calculation unit 58 takes the difference between the time stamp information included in the packet and the time stamp information when the previous packet arrived, and transmits the packet. Calculate the interval Ts. At the same time, the packet reception interval calculation unit 61 extracts a time stamp † blueprint counting the clock frequency from its own frequency variable clock source 59 by the time stamp extraction unit 60, and obtains a difference from the previous time stamp extraction time. Thus, the packet reception interval Tr is calculated. The arithmetic correction unit 62 uses the packet transmission interval Tr and the packet reception interval Ts to calculate a frequency variable clock source based on a value calculated by an arithmetic expression of a * (Tr Ts) ZTr ( _a is a predetermined constant). 59 to control the clock frequency of the slave device 52 to match the clock frequency of the master device 51.

As described above, the conventional clock synchronization system performs clock synchronization for the purpose of matching the phases of the transmission device and the reception device and preventing the audio buffer of the reception device from failing.

[0009] Further, Japanese Patent Application Laid-Open No. 2004-153546 discloses a clock synchronization method for matching the clock signal frequency of one terminal device with the clock signal frequency of another terminal device. In the method described in this document, the frequency variable clock source is controlled based on the calculated differential force between the packet transmission interval of the master device and the packet reception interval of the slave device. Thus, the clock signal frequencies can be easily matched.

However, the following problems exist when performing the clock synchronization method in the audio transmission system of FIG. 12 described above.

[0011] The above-described conventional technique is a technique for matching the phases of the transmission side and the reception side. In the prior art, the clock frequency is controlled based on the difference between the packet interval on the transmitting side and the packet interval on the receiving side, and the clock frequencies on the transmitting side and the receiving side are matched. Thus, in the prior art, the phases of the transmission side and the reception side can be matched. However, the conventional clock synchronization system does not realize clock synchronization that accurately matches the audio output times between a plurality of receiving devices. For this reason, there is a problem that a difference in audio output time may occur between a plurality of receiving apparatuses. This problem will be described in more detail below.

As a precondition for controlling the frequency variable clock source, there is a restriction that the control value for the frequency variable clock source is smaller than the resolution allowed by the oscillator and cannot be set to a unit value. To do. For this reason, every time the frequency variable clock source is controlled, an error of a unit size less than the resolution occurs between the value that is originally set and the value that is actually set after truncation by the resolution.

[0013] This error becomes a problem in the prior art. In the prior art, the frequency variable clock source is controlled based on the value calculated from the difference between the packet transmission interval of the master device and the packet reception interval of the slave device. Therefore, an error below the above resolution occurs every time a knot is received, and this error is accumulated. If errors are accumulated, the total time stamp value from the start of synchronization on the slave side will deviate from the total time stamp value on the master side. Such a total time stamp difference cannot be determined by a conventional control method using a packet interval. Since errors are accumulated in each of the plurality of slave devices, the total time stamp value is shifted among the plurality of slave devices. As a result, there was a problem that the expansion and contraction of the audio delay time existed in multiple slave devices. Furthermore, there is a problem that the difference in delay time causes a difference in audio output time between a plurality of slave devices.

Disclosure of the invention

Problems to be solved by the invention [0014] The present invention has been made under the above-described background, and its object is to reduce the expansion and contraction of the delay time of the sound and to minimize the difference in the sound output time between the receiving devices. It is to provide a clock synchronization system and method.

[0015] It is another object of the present invention to provide a clock synchronization system and method capable of minimizing a difference in audio output time between receiving apparatuses even in a network environment having disturbance elements.

Means for solving the problem

The clock synchronization system of the present invention is a system that uses a synchronization packet to perform clock synchronization between a main apparatus that performs audio transmission via a network and a remote apparatus, and the main apparatus includes a clock source A time stamp generating unit that generates time stamp information and a packet transmitting unit that transmits a synchronization packet in which the time stamp information is embedded to the network, and the remote device includes a packet receiving unit that receives the synchronization packet from the network, The main unit cumulative time stamp calculation unit that calculates the main unit total type stamp value To, which is the total time stamp value of the synchronization start force, from the time stamp information included in the synchronization bucket, and a clock capable of variable frequency control Source time stamp information is extracted, and the total time of remote device is the time stamp value of the synchronous start force. The remote device cumulative time stamp calculation unit that calculates the time stamp value Ti and the frequency of the clock source capable of variable frequency control are corrected based on the difference between the main device cumulative time stamp value To and the remote device cumulative time stamp value Ti. A correction calculation unit.

[0017] Another aspect of the present invention is a clock synchronization method that uses a synchronization packet to synchronize clocks between a main device and a remote device that perform voice transmission via a network. The master unit generates the time stamp information to generate the clock source time stamp information, and transmits the synchronization packet in which the time stamp information is embedded to the network, and the remote unit receives the synchronization packet from the network and is included in the synchronization packet. Calculates the main device cumulative type stamp value To, which is the cumulative time stamp value of the synchronization start force from the time stamp information that is generated, extracts the clock source power time stamp information that allows variable frequency control, and accumulates from the start of synchronization. The remote device cumulative time stamp value Ti, which is the time stamp value of the clock, is calculated and the frequency of the clock source capable of variable frequency control is calculated. The number, and the main apparatus cumulative time stamp value To Correct according to the difference of the remote device cumulative time stamp value Ti.

[0018] Another aspect of the present invention is a remote device that is provided in a system that performs voice transmission via a network and that performs clock synchronization using a synchronization packet that has also received the main device power. The packet receiving unit that receives the synchronization packet in which the time stamp information generated from the clock source of the device is embedded, and the time stamp information included in the synchronization packet are the time stamp values accumulated from the start of synchronization. The main unit cumulative time stamp calculation unit that calculates the unit total type stamp value To and the clock source power capable of variable frequency control on the remote unit side Extract the time stamp information and the remote time stamp value is the cumulative time stamp value of the synchronization start power Remote device cumulative time stamp calculation unit that calculates device cumulative time stamp value Ti, and a clock that can be controlled with variable frequency. The frequency of the source, and a, a correction operation unit for correcting, based on the difference of the main apparatus cumulative timestamp value To and the remote device cumulative time stamp value Ti.

As described below, there are other aspects of the present invention. Accordingly, this disclosure is intended to provide some aspects of the invention and is not intended to limit the scope of the invention described and claimed herein.

Brief Description of Drawings

FIG. 1 is a block diagram showing a clock synchronization system according to an embodiment of the present invention. FIG. 2 is a block diagram of an audio transmission system equipped with a clock synchronization system.

[Figure 3A] Figure 3A is a diagram for explaining the outline of the Haas effect.

[Figure 3B] Figure 3B is a diagram for explaining the outline of the Haas effect.

[FIG. 4] FIG. 4 is a diagram showing an example in which the audio transmission system is applied to evacuation guidance broadcasting.

[FIG. 5] FIG. 5 is a diagram showing an embodiment in which a Dolby Digital 5.1 channel acoustic broadcast is broadcast from a distant place to multiple points in an audio transmission system.

[Fig. 6] Fig. 6 is a diagram for showing the transition of the time stamp value in the clock synchronization system. [Fig. 7] Fig. 7 is a diagram for showing an overview of synchronization packet transmission and reception in the clock synchronization system.

[Fig.8] Fig.8 is a diagram to show the setting error of the frequency variable clock source in the clock synchronization system [Fig. 9A] Fig. 9A is a diagram to show the error when clock synchronization is performed.

[Fig. 9B] Fig. 9B is a diagram for showing the time stamp value transition when clock synchronization is performed. [Fig. 10] Fig. 10 is a diagram for showing an outline of processing when a packet including a delay arrives. [Fig. 11] Fig. 11 is a diagram for showing an overview of the processing when the timestamp value difference exceeds the threshold.

[FIG. 12] FIG. 12 is a diagram for showing a conventional clock synchronization system.

Explanation of symbols

[0020] 10 clock synchronization system (synchronization means)

11 Master device

12 Slave device

13, 34, 35 network

31 sound source

32 Speaker

33 Audio transceiver

41 Voice delivery unit

42 Audio reception unit

44 Dolby Digital 5. 1 channel space

BEST MODE FOR CARRYING OUT THE INVENTION

[0021] A detailed description of the present invention will be described below. However, the following detailed description and the accompanying drawings do not limit the invention. Instead, the scope of the invention is defined by the appended claims.

The clock synchronization system of the present invention is a system that uses a synchronization packet to perform clock synchronization between a main apparatus that performs audio transmission via a network and a remote apparatus, and the main apparatus includes a clock source A time stamp generating unit that generates time stamp information and a packet transmitting unit that transmits a synchronization packet in which the time stamp information is embedded to the network, and the remote device includes a packet receiving unit that receives the synchronization packet from the network, A master unit cumulative time stamp calculation unit that calculates a master unit cumulative type stamp value To, which is a cumulative time stamp value of the synchronization start force, from the time stamp information included in the synchronization bucket; Clock source power capable of variable wave number control Time stamp information is extracted, and the remote device cumulative time stamp calculation unit for calculating the remote device cumulative time stamp value Ti, which is the time stamp value of the synchronous start power, and frequency variable control And a correction calculation unit that corrects the frequency of the clock source capable of being corrected based on the difference between the main device accumulated time stamp value To and the remote device accumulated time stamp value Ti.

[0023] With this configuration, it is possible to control the frequency of the clock source while monitoring the distance of the total time stamp by referring to the time stamp (To, Ti) of the synchronization start time. Is done. Therefore, for errors that occur each time the frequency variable clock source is controlled (that is, errors that occur due to the difference between the value that you originally want to set and the value that is actually set after truncation by the resolution), the error is accumulated. Never go. Only the error for one packet that arrives is included in the control. This one-packet error has little effect on the total timestamp difference. In this way, the difference between the time stamp values of the master device and the slave device can be minimized, and the difference in the audio output time between the slave devices can be minimized.

[0024] Further, the correction calculation unit obtains the control value P for controlling the frequency of the clock source capable of variable frequency control as the difference between the main device cumulative time stamp value To and the remote device cumulative time stamp value Ti (To-Ti ) And the difference (To-Ti) may be determined based on the degree of continuous tilting in one direction.

[0025] Here, "tilt" means that the difference (To-Ti) deviates from 0, that is, becomes positive or negative. In addition, “tilt” means that Ti is larger or smaller than To. The degree of tilt is typically expressed as the number of tilts (number of consecutive tilts) as follows.

With this configuration, the clock frequency is controlled based on the degree to which the time stamp difference is tilted in one direction. As a result, when a difference occurs, it is possible to operate a control that pulls back and reduces the difference more rapidly in the direction opposite to the generated difference. For this reason, it is possible to prevent the time stamp difference from spreading greatly, and to stably keep the audio output time difference between slave devices to a minimum.

[0027] Further, the correction calculation unit determines the control value P according to the calculation formula of P = a * (To-Ti) + c. The wiggle constant a is a fixed constant and the variability coefficient c is a value that fluctuates in proportion to the degree to which the difference (To-Ti) is continuously tilted in one direction.

[0028] With this configuration, the variation coefficient c is used, and the variation coefficient c is changed in proportion to the degree that the time stamp difference is tilted in one direction. As a result, when a difference occurs, it is possible to operate a control for pulling back the difference more rapidly in the direction opposite to the generated difference. For this reason, it is possible to prevent the time stamp difference from spreading greatly, and to stably keep the audio output time difference between slave devices to a minimum. The value of constant a used at this time is a fixed constant. If the constant a is set to a large value, the time stamp difference will be controlled to vibrate up and down densely from 0 as time passes. If the constant a is set to a small value, the time stamp value difference gently swings up and down at 0. The constant a is set in consideration of such a phenomenon, and suitable control according to the time stamp difference can be performed using the variation coefficient c as described above.

[0029] Further, the correction calculation unit may determine the variation coefficient c according to the number of times the difference (To-Ti) continuously tilts. More specifically, the correction calculation unit determines that the coefficient of variation c is c = (previous c) + k when (To-Ti)> 0 continues N times, and (To-Ti) <0 It may be configured to define c = (previous c) −k when N times in succession. Here, N is a predetermined number of times and k may be a predetermined constant.

[0030] By using such a variation coefficient c, for example, when N = 3, if the difference in time stamps becomes To> Ti three times in succession, the value of c is increased. As a result, the value P for controlling the frequency of the clock source is set to a large value, the oscillation frequency on the slave device side is set to a large value, and Ti is controlled so as to approach To earlier. On the other hand, if the time stamp difference is To and Ti for 3 consecutive times, the value of c is decreased. As a result, the value P for controlling the frequency of the clock source is set small, the oscillation frequency on the slave device side is set small, and Ti is controlled so as to approach To earlier. By these methods, the time difference of the time stamp ^ can be detected earlier, and the difference can be pulled back and reduced. As for the value of k used here, the greater the value k, the greater the control that pulls back the time stamp difference in the reverse direction, but the difference when tilted in the reverse direction also increases. . On the other hand, the smaller the value k, the difference in timestamps over time. The control works to slowly pull the minute in the reverse direction, but the difference when tilting in the reverse direction can be reduced. Considering such a phenomenon, the value k is set, and a suitable control according to the time stamp difference can be performed using the coefficient of variation C as described above.

[0031] The correction calculation unit may be configured to determine the coefficient of variation c as c = (P set last time) when (To-Ti) = 0.

[0032] Thus, when (To-Ti) = 0, the time stamp values of the master device and the slave device are substantially the same. In such a state, the correction calculation unit does not change the value P that controls the frequency of the clock source set last time. As a result, the synchronization establishment state can be maintained for a while. However, even if this condition continues, it is unlikely that the frequency of the master device and the frequency of the slave device will be exactly the same over a long period of time. Therefore, the difference should occur in either direction. In that case, the clock synchronization control shown in the present invention is again performed in the same manner, and the frequencies of the master device and the slave device can be matched. In this way, suitable synchronization control between the master device and the slave device can be performed.

[0033] Further, the main apparatus may be configured to transmit the synchronization packet by IP multicast or IP broadcast.

[0034] With this configuration, even when there are a large number of slave devices in the system, by transmitting synchronization packets by IP multicast or IP broadcast, it is easy to keep all the loads on the network constant. Slave devices can receive synchronization packets. Even when a new slave device is added to the system, clock synchronization can be started while keeping the network traffic constant, especially without increasing the new load.

[0035] In addition, the main device transmits the synchronization packet at regular intervals, and the remote device monitors the estimated arrival time of the synchronization packet received from the main device, and exceeds the predetermined range from the estimated packet arrival time. Remove the synchronization packets that arrived from the target of the processing that controls the frequency of the clock source that can be controlled with variable frequency.

Here, exceeding the predetermined range means that the synchronization packet arrives before or after the predetermined range. To wear.

[0037] With this configuration, the scheduled arrival time of the next packet is monitored, and packets that arrive outside the allowable range W are invalidated. Therefore, packets containing large arrival delays can be eliminated. Synchronous control is not performed for invalidated packets, but can be easily corrected by normal packets that arrive next time. Therefore, there is no significant effect on clock synchronization. By referencing only accurate time stamps in this way, it is possible to prevent a decrease in synchronization accuracy due to disturbance factors such as a network.

[0038] When the difference (To-Ti) exceeds a predetermined threshold value R, the remote device resets the accumulated cumulative time stamp value To and the accumulated remote time stamp value Ti that have been accumulated so far. The accumulation of time stamp values may be resumed.

[0039] Of course, the above-described reset processing may be performed when the difference (To-Ti) increases to either the positive or negative side and exceeds the threshold value.

[0040] This configuration provides the following advantages. If the reference time stamp value acquired immediately after the start of synchronization contains a large delay, the slave device performs synchronization control using the incorrect value, and there is a problem that normal synchronization control cannot be performed. In order to solve this problem, it is preferable to adopt the above configuration. With the above configuration, even when the reference time stamp value acquired for the first time includes a delay exceeding the threshold, if the normal time stamp value is received even once, the difference value exceeds the threshold. Then, a new accurate time stamp value can be acquired again. In this way, it becomes possible to further enhance the tolerance to delays caused by disturbance factors such as the network.

[0041] As described above, according to the present invention, it is possible to reduce the audio output time difference between slave devices.

[0042] Further, even if the number of slave devices existing in the system increases, clock synchronization can be performed without changing the load of network traffic.

[0043] Furthermore, even when a wide area network such as a WAN is used, the clock synchronization system with high clock synchronization accuracy can be obtained by eliminating the bucket information including large fluctuations and always referring to only the accurate time stamp value. Can be provided. Hereinafter, a clock synchronization system according to an embodiment of the present invention will be described with reference to the drawings. The clock synchronization system of the present embodiment is provided in an audio transmission system as shown in FIG. 2, and realizes clock synchronization between audio transmission / reception devices 33.

The audio transmission system in FIG. 2 includes a sound source 31, a speaker 32, an audio transmission / reception device 33, a local LAN 34, and a WAN 35. The voice transmission / reception device 33 is a device capable of sending and receiving voice data via the local LAN 34 and WAN 35, and has a function of packetizing voice data in order to stream the voice data and transmit it over the network. . In addition, the sound source 31 is connected to the sound transmitting / receiving device 33 as a signal source, and the speaker 32 is connected to the sound transmitting / receiving device 33 as sound emitting means.

[0046] As an application of such a system, multipoint broadcasting is mainly assumed. For this broadcast, voice communication is realized in buildings such as office buildings, multi-store commercial facilities, stations, and airports. As the sound source 31, for example, a CD player, an MD player, an IC player, a remote control microphone, etc. for broadcasting BGM and emergency broadcast sound source are connected. As the speaker 32, for example, various types of speakers such as a ceiling-embedded speaker and a hanging speaker are connected. In addition, the configuration via the WAN 35 enables voice transmission between different LANs via a wide area network such as the Internet.

[0047] In addition to the above-mentioned broadcasting use, by realizing sound field control using the Haas effect on the speaker 32 side, it is possible to provide a more sophisticated service.

FIG. 3A and FIG. 3B are diagrams for explaining the phenomenon of the Haas effect. The Haas effect is a phenomenon that gives a sense of direction to sound using the characteristics of human hearing. When sound is output from multiple speakers, humans recognize that the sound is coming in the direction of the speaker that emitted the first sound. When the same sound is being output from two speakers with the same receiving point power, if one speaker is delayed, humans can hear the direction force of the other speaker. To recognize. In the example of Fig. 3A, if speakers A and B emit sound at the same time, humans recognize that the sound is being heard from nearby speaker B. However, as shown in Figure 3B, if a delay is applied so that the sound from nearby speaker B arrives later than far away speaker A, humans recognize that far away speaker A is sounding. The

[0049] A service using sound field control can be provided by intentionally controlling the direction of sound using this phenomenon. Specifically, various services such as evacuation guidance broadcasting in office buildings, acoustic design in theaters and concert halls, and Dolby Digital 5.1 channel acoustic broadcasting can be realized via the network.

[0050] FIG. 4 is a diagram showing an embodiment using evacuation guidance broadcasting in an office building. An audio transmission unit 41 is arranged on the broadcast room side, and a plurality of audio reception units 42 are arranged on the reception side. The voice transmission unit 41 and the plurality of voice reception units 42 are arranged via the campus lan 34. The audio transmission unit 41 includes a sound source and an audio transmission device. The sound receiving unit 42 includes a sound receiving device and a speaker provided on the aisle ceiling of the building.

[0051] Now, when evacuation guidance broadcasting is started, delay times are individually set for a plurality of audio receiving units 42 (speakers). This allows humans to recognize when they hear the direction force sound they want to evacuate. In this case, the audio streams that are also output by each speaker are data having the same content. However, the output delay time is different among multiple speakers. In order to realize evacuation guidance, it is necessary to set the delay time in the direction of evacuation to be small and to deliver the sound of the speaker in that direction to the ear of human earliest.

FIG. 5 is a diagram showing an embodiment in which Dolby Digital 5.1 channel sound broadcasting is broadcast to a plurality of remote locations. An audio transmission unit 41 is arranged on the broadcast room side, and the audio transmission unit 41 includes a sound source and an audio transmission / reception device. In addition, a plurality of Dolby Digital 5.1 channel spaces 44 are arranged via the local area LAN 34. In each 5.1 channel space, there are multiple speakers corresponding to 5.1 channels. Each speaker is equipped with a voice receiver. In the case of Dolby Digital 5.1 channel, unlike evacuation-guided broadcasting, the same stream is broadcasted in the same 5.1 channel space to provide sound effects used in movies, etc. There are many opportunities to broadcast independent audio on the channel! ,. [0053] In order to realize these services via a network, even when audio data is transmitted over a network with disturbance factors, the original audio output time difference existing between the speakers is minimized. There is a need. The original audio output time difference is the time difference when performance of the hearth effect is not emphasized (performance value). If the original audio output time difference is not small, the Haas effect will not occur well when delay is applied. Specifically, it is said that the limit value of the audio output time difference that can make effective use of the Haas effect when the installation interval of speakers is several meters to several tens of meters is 2 milliseconds. In order to minimize the original audio output time difference, clock synchronization must be accurately realized between the transmitting and receiving devices in the audio transmission system.

Hereinafter, a clock synchronization system according to an embodiment of the present invention will be described with reference to the drawings. The purpose of this system is to enable the provision of the above services. Specifically, this system can be suitably applied to services using the Haas effect. In addition, this system can be suitably applied to sound field control such as 5.1 channel broadcasting.

FIG. 1 is a block diagram showing a configuration of clock synchronization system 10 according to the present exemplary embodiment. As shown in FIG. 1, the clock synchronization system 10 includes a master device 11 and a slave device 12, and the master device 11 and the slave device 12 are connected via a network 13.

[0056] The clock synchronization system 10 in FIG. 1 is realized by the configuration of the audio transmission system in FIG. The master device 11 is composed of one of the plurality of audio transmission / reception devices 33 in FIG. The slave device 12 is also composed of one of the plurality of audio transmission / reception devices 33 in FIG. The network 13 corresponds to the LAN 34 and the WAN 35 in FIG.

In FIG. 1, one slave device 12 is shown. In practice, a plurality of slave devices 12 may be provided. That is, each of the plurality of voice transmitting / receiving devices 33 in FIG. The plurality of slave devices 12 have the same function and realize clock synchronization by the same operation. Therefore, in the following description, the clock synchronization system 10 of the present embodiment will be described by focusing on one slave device 12.

Further, the master device 11 and the slave device 12 in FIG. 1 are respectively the main device of the present invention. And respond to remote devices.

In FIG. 1, as the master device 11, one device is set in advance in the system. The master device 11 has a function of transmitting a packet containing its own time stamp information to the slave device 12 via the network 13.

The clock synchronization system 10 according to the present embodiment is not a type that adds time stamp information for clock synchronization to an audio stream. In the clock synchronization system 10, one master device 11 existing in the system transmits its own clock synchronization packet. This is because one device considers the case where there are multiple audio channels. Suppose that the slave device that owns multiple channels performs clock synchronization for each time stamp information embedded in separate audio streams. In this case, it is necessary to configure the number of audio channels for all systems including the processor, so the configuration is effective. Although the clock of one set master device 11 is used, it can be said that a configuration in which all the slave devices 12 and all the audio channels existing in the device are synchronized is effective. In addition, mounting multiple audio channels in one device can reduce the number of devices existing in the system, and can provide an effective configuration from the viewpoint of cost and system construction. From this point of view, synchronous packets are advantageously used.

[0061] In addition, since the clock synchronization of the present embodiment aims to make the entire system subordinate to one clock, the master device 11 is not necessarily limited to the voice transmission device. In the audio transmission system 10, all the devices other than the master device 11 become the slave device 12. Therefore, one or more slave devices 12 exist in the system and have a function of receiving a synchronization packet from the master device 11 via the network 13. As the network 13, for example, a LAN or WAN capable of IP protocol communication can be used.

The master device 11 includes a clock source 14, a time stamp extraction unit 15, and a packet transmission unit 16. The clock source 14 is composed of an element capable of generating a system reference clock such as a crystal oscillator. The time stamp extraction unit 15 includes an interface for extracting clocks of 14 clock sources, and a time stamp that counts the frequency power of the obtained clocks. And an interface for discharging the packet information to the packet transmission unit 16 at the subsequent stage. The packet sending unit 16 includes an interface for obtaining the time stamp information from the time stamp extracting unit 15 and an interface for packetizing the information and discharging it onto the network 13. As a packet method, for example, communication based on the IP protocol can be performed. It is also preferable to implement a protocol that can use IP multicast. IP broadcast may also be used. This makes it possible to receive packets by multiple specific slave devices while keeping the load on the network constant.

On the other hand, the slave device 12 includes a packet receiving unit 17, a master device reference time stamp storage unit 18, a master device cumulative time stamp calculation unit 19, a frequency variable clock source 20, and a slave device reference time stamp storage unit 21. The slave device cumulative time stamp calculation unit 22 and the calculation correction unit 23 are provided.

The packet receiving unit 17 includes an interface that receives a clock synchronization packet from the network 13. The packet receiving unit 17 uses, for example, an interface having a protocol capable of receiving an IP multicast packet. As a result, the plurality of slave devices 12 can receive the clock synchronization packet from the master device 11 with a constant load. A packet may be received by IP broadcast.

The master device reference time stamp storage unit 18 is configured by a memory that stores the time stamp information of the master device received at the first time when clock synchronization starts as a reference value. This time stamp value becomes reference information for calculating the accumulated time stamp value of the master device when performing clock synchronization thereafter.

[0066] The master device cumulative time stamp calculation unit 19 refers to the master device time stamp information received from the second time onward, and the reference time stamp information at the start of clock synchronization obtained from the master device reference time stamp storage unit 18. Then, it has an interface for calculating the difference between the two stamp information, and thereby discharging the total time stamp of the master device from the start of clock synchronization to the arithmetic correction unit 23.

[0067] The frequency variable clock source 20 is a clock source in the slave device 12, and is composed of an element capable of changing the clock frequency. The clock source 20 is, for example, a voltage It can be composed of a crystal oscillator whose frequency can be changed by fluctuation.

The slave device reference time stamp storage unit 21 is configured by a memory that stores, as a reference value, time stamp information of the slave device when clock synchronization starts and a synchronization packet is received for the first time. This time stamp value is used as reference information for calculating the accumulated time stamp value of the slave device in the subsequent clock synchronization.

[0069] The slave device cumulative time stamp calculation unit 22 refers to the time stamp information of the slave device received after the second time and the reference time stamp information at the start of clock synchronization obtained from the slave device reference time stamp storage unit 21. Then, it has an interface for calculating the difference between the time stamp information and discharging the total time stamp of the slave device having the power at the start of clock synchronization to the arithmetic correction unit 23.

[0070] The calculation correction unit 23 extracts an interface for extracting the time stamp value To from the master device cumulative time stamp calculation unit 21 and a time stamp value Ti from the slave device total time stamp calculation unit 22 Has an interface. The arithmetic correction unit 23 adjusts the frequency of the frequency variable clock source 20 based on the integer value. Then, the arithmetic correction unit 20 controls the frequency variable clock source 20 based on the value calculated by the difference value (To-Ti) between the two time stamp values, thereby obtaining the cumulative time stamp value of the slave device 12. To the cumulative time stamp value of master device 11.

In this manner, the clock synchronization system according to the present embodiment can minimize the difference between the accumulated time stamp values of the master device and the slave device. The audio output time difference between the two can always be minimized. The cumulative time stamp value of the master device and the cumulative time stamp value of the slave device correspond to the main device cumulative time stamp value To and the remote device cumulative time stamp value Ti of the present invention, respectively.

Hereinafter, the operation of the embodiment will be described. Here, the operation of the slave device 12 depending on the clock of the master device 11 via the network 13 will be described with reference to the respective drawings.

[0073] Figure 6 shows how to keep the audio output time difference at a minimum. Thus, it is effective to control the clock source of the slave device 12 so that the accumulated time stamp values of the master device 11 and the slave device 12 are always close to each other. Such control is realized by the clock synchronization system 10 and its clock synchronization method of the present embodiment.

When the slave device 12 receives a synchronization packet for the first time, it is stored in the master device reference time stamp storage unit 18 as a reference value for clock synchronization control performed after the time stamp information included in the packet. At the same time, the slave device 10 stores the reference time stamp value acquired from the frequency variable clock source 20 of the slave device 10 in the slave device reference time stamp storage unit 21. Thereafter, since the master device 11 transmits the synchronization packet at regular intervals, the slave device 12 also periodically receives the synchronization packet.

[0075] However, packet arrival may be delayed more than usual due to the occurrence of disturbance factors such as a network. In consideration of this, as shown in FIG. 10, when the packet arrives, the slave device 12 monitors the estimated arrival time of the next packet as well. Then, the slave device 12 determines whether or not the corresponding next packet arrives outside the allowable range W. In the present embodiment, as shown in the figure, the allowable range is determined based on the estimated arrival time of the next packet, and specifically, is set to the estimated arrival time ± W. Then, the slave device 12 invalidates the packet that has arrived outside the allowable range W (this corresponds to the case where the packet has arrived beyond the predetermined range to the front side or the rear side). On the other hand, packets arriving within the allowable range W are determined not to affect the voice output reception difference. This normal and valid packet is processed as a target of clock synchronization control.

[0076] By such packet invalidation processing, synchronization control is not performed for a packet that is invalidated. However, since it can be easily corrected by normal packets that arrive next time, there is almost no effect of packet invalidation.

[0077] With the above processing, a packet that has arrived normally is subject to clock synchronization control. here

The cumulative time stamp values To (master device) and Ti (slave device) for the initial packet reception are expressed as shown in Fig. 7 on the time axis.

The arithmetic correction unit 23 of the slave device 12 controls the oscillation frequency by performing voltage control on the frequency variable clock source 20. In the present embodiment, the calculation correction unit 23 is the master The oscillation frequency is controlled according to the difference between the cumulative time stamp value To of one device 11 and the cumulative time stamp value Ti of the slave device 12. The correction calculation unit 23 is configured to adjust the frequency based on the integer value. By this control, the cumulative time stamp value Ti of the slave device 12 is brought close to the cumulative time stamp value To of the master device 11.

More specifically, the calculation correction unit 23 performs voltage control from the control value P calculated by the following calculation formula. a is a fixed constant (coefficient), and c is a coefficient of variation described later.

[0080] P = a water (To—Ti) + c

Here, FIG. 8 shows an error when controlling the frequency variable clock source 20.

As shown in FIG. 8, the frequency variable clock source 20 has a set resolution. Therefore, it must be noted that the control of the frequency clock source 20 causes an error due to the resolution. This error is an error due to the difference between the value that is originally set and the value that is actually set to the frequency variable clock source. This error occurs every time each packet is processed. In the conventional technology, this error accumulates, and the accumulated stamp value greatly deviates between the master device and the slave device. Also, the accumulated error differs between slave devices, and the accumulated time stamp value also deviates between slaves. As a result, the accuracy of clock synchronization decreases, and the difference in audio output time between slave devices increases.

However, in the present embodiment, control is performed with reference to the accumulated time stamp value from the start of synchronization. Therefore, as shown in Fig. 9A, even if an error occurs, only the error for one packet that arrives at that time is included. This error has little effect on the difference in cumulative timestamp values. Accordingly, the cumulative type stamp values of the master device 11 and each slave device 12 can be brought close to each other. As a result, the time stamp values can be made closer between the plurality of slave devices 12.

[0083] The constant a is a fixed constant. When the constant a is set to a large value, the time stamp difference is controlled to vibrate up and down densely from 0 as time passes. In addition, if the value of the constant a is set to a small value, the control that the vertical difference of the time stamp value gently oscillates with 0 as the boundary works. In consideration of such a phenomenon, the constant a is set, and thus it is possible to perform suitable control according to the time stamp difference.

[0084] Further, as shown in the above arithmetic expression, in the present embodiment, the variation coefficient c is a frequency control. Used for control. The coefficient of variation c is added to a * (To—Ti) (a value corresponding to the difference between the time stamp values), whereby the control value P is calculated. The change factor c changes in proportion to the degree to which the time stamp difference is tilted in one direction. By changing the value of the coefficient of variation c in this way, as shown in FIG. 9B, it is possible to obtain the effect of rapidly pulling back the generated difference in the reverse direction and reducing it, and the time stamp difference is greatly expanded. Can be prevented. For this reason, it is possible to prevent the time stamp difference from spreading greatly, and to stably keep the audio output time difference between the slave devices to a minimum.

[0085] As described above, the variation coefficient c changes in proportion to the degree to which the time stamp difference is tilted in one direction. “Tilt” means that the difference (To—Ti) deviates from 0, that is, it becomes positive or negative. “Tilt” means that the force vj, where Ti is larger than To, It is also a difference between the two. The degree of tilt is typically represented by the number of tilts (number of consecutive tilts) as follows. In this case, the correction calculation unit 23 determines the variation coefficient c according to the number of times the difference (To-Ti) is continuously tilted. More specifically, the correction calculation unit 23 determines the coefficient of variation c as c = (previous c) + k when (To−Ti)> 0 continues N times. In addition, the correction calculation unit 23 determines the coefficient of variation c as c = (previous c) k when (To—Ti) く 0 continues N times. Here, N is a predetermined number of times, and k is a predetermined constant.

For example, assume that N = 3. If the difference in time stamps is To> Ti for three consecutive times, the value P that controls the frequency of the clock source by increasing the value of c is set to a large value, and the oscillation frequency on the slave device side is set to a large value. And Ti is controlled to get closer to To sooner. On the other hand, if the difference in time stamps becomes To and Ti for 3 consecutive times, the value P that controls the frequency of the clock source is reduced by decreasing the value of c, and the oscillation frequency on the slave device side is reduced. It is set small and Ti is controlled to get closer to To sooner. By using these methods, the spread of the time stamp difference can be detected more quickly, and the difference can be pulled back in the direction of decreasing the difference.

[0087] Regarding the value of k used here, the larger the value k, the faster the time stamp difference is pulled back in the reverse direction, but the difference when tilted in the reverse direction is also larger accordingly. Become. On the other hand, as the value k is smaller, control is performed so that the time stamp difference is gradually pulled back in the reverse direction over time, but the difference when tilted in the reverse direction is reduced. Is possible. Considering this phenomenon, the value k is set appropriately. As described above, it is possible to perform suitable control according to the time stamp difference using the coefficient of variation C.

Further, the correction calculation unit 23 determines the variation coefficient c as c = (P set last time) when (To-Ti) = 0. By this processing, synchronous control is performed as follows.

[0089] When (To-Ti) = 0, the time stamp values of the master device and the slave device are almost the same. In such a state, the correction calculation unit 23 does not change the value P that controls the frequency of the previously set clock source. As a result, the synchronization establishment state can be maintained for a while. However, even if this condition continues, it is unlikely that the frequency of the master device and the frequency of the slave device will be exactly the same over a long period of time. Therefore, the difference should occur in either direction. In that case, the clock synchronization control shown in the present invention is again performed in the same manner, and the frequencies of the master device and the slave device can be matched. In this way, suitable synchronization control between the master device and the slave device can be performed.

As described above, master device 11 transmits a synchronization packet, slave device 12 receives the synchronization packet, and performs synchronization control on the clock source. By repeating this operation, the master device 11 and the slave device 12 adjust the cumulative time stamp value.

In the above embodiment, the master device 11 and the slave device 12 are in a one-to-one relationship. However, the actual configuration may be one-to-many. The other slave devices 12 have the same configuration and perform the same operation. In this way, it is possible to maintain the audio output time difference between the plurality of slave devices 12 to a minimum.

Further, FIG. 11 is a diagram showing a coping method when a large delay is included in the packet received by the slave device 12 for the first time. Assume that the standard time stamp value acquired for the first time includes a large delay. In this case, the system continues synchronization control using the incorrect reference timestamp value. In order to solve this point, a predetermined threshold R is set in this embodiment. When the time stamp difference (To—Ti) exceeds the threshold value R, the slave device 12 accumulates the accumulated time status up to that point. Initialize the amplifier value and resume accumulating a new cumulative timestamp value. More specifically, the accumulated time stamp value To of the master device 11 and the accumulated time stamp value Ti of the slave device 12 are reset, and accumulation of the new time stamp value is resumed.

This configuration provides the following advantages. Suppose that the first time stamp value that is acquired for the first time includes a delay exceeding the threshold value R. In this case, when a normal time stamp value arrives after that, the difference value (To-Ti) exceeds the threshold value, and a new accurate time stamp value can be obtained again. In this way, it is possible to further enhance resistance to delay caused by disturbance elements such as a network.

[0094] It should be noted that the above processing may be performed when the difference value (To-Ti) exceeds the threshold value on either the positive or negative side. More specifically, a positive threshold value and a negative threshold value are set, and the above process may be performed when the difference value exceeds either threshold value. Also, a positive threshold value may be set, and the absolute value of the difference value may be compared with the threshold value.

As described above, by using the clock synchronization system or method according to the present embodiment, it is possible to construct an audio transmission system with no audio output time difference between slave devices.

[0096] Even if the number of slave devices existing in the system increases, clock synchronization can be performed without changing the load of network traffic.

[0097] Further, even when a wide area network such as a WAN is used, the bucket information including large fluctuations is eliminated, and only an accurate time stamp value is always referred to, thereby enabling a voice transmission system with high clock synchronization accuracy. Can be provided.

[0098] Then, the clock synchronization system and the audio transmission system including the clock synchronization system according to the present embodiment can be used for broadcasting to remote places and multi-point broadcasting systems such as multi-store commercial facilities! / Applicable to. In addition, this embodiment is a service that uses sound field control via a network, such as evacuation guidance broadcasting in office buildings, acoustic design in theaters and concert halls, and Dolby Digital 5.1 channel acoustic broadcasting. Can be applied to.

[0099] The power of explaining the preferred embodiment of the present invention considered at the present time. It will be understood that various modifications can be made to the present embodiment, and the true spirit and scope of the present invention can be understood. It is intended that the appended claims include all such variations that fall within the scope.

Industrial applicability

As described above, the clock synchronization system according to the present invention can increase the clock synchronization accuracy, and is useful as a clock synchronization system such as an audio transmission system that performs sound field control or the like.

Claims

The scope of the claims

[1] A clock synchronization system that uses synchronous packets to synchronize clocks between a main device and a remote device that perform voice transmission over a network.

The main unit is

Clock source time stamp generation unit for generating time stamp information;

A packet transmission unit that transmits the synchronization packet in which the time stamp information is embedded to the network;

The remote device is:

A packet receiver that receives the synchronization packet; and

A main device cumulative time stamp calculating unit that calculates a main device cumulative type stamp value To that is a cumulative time stamp value of the synchronization start power from the time stamp information included in the synchronization packet;

A remote device cumulative time stamp calculating unit for extracting clock source power time stamp information capable of variable frequency control on the remote device side and calculating a remote device cumulative time stamp value Ti that is a cumulative time stamp value from the start of synchronization;

A correction arithmetic unit for correcting the frequency of the clock source capable of variable frequency control based on a difference between the main device accumulated time stamp value To and the remote device accumulated time stamp value Ti;

A clock synchronization system comprising:

[2] The correction calculation unit controls the control value P for controlling the frequency of the clock source capable of variable frequency control as a difference between the main device accumulated time stamp value To and the remote device accumulated time stamp value Ti (To— 2. The clock synchronization system according to claim 1, wherein the clock synchronization system is determined based on a degree of Ti) and the difference (To-Ti) continuously tilted in one direction.

[3] The correction calculation unit determines the control value P according to a calculation formula of P = a * (To—Ti) + c, where the constant a is a fixed constant and the variation coefficient c is the difference ( 3. The clock synchronization system according to claim 2, wherein To—Ti) is a value that varies in proportion to the degree of continuous tilting in one direction.

[4] The correction calculation unit tilts the variation coefficient c continuously with the difference (To-Ti). 4. The clock synchronization system according to claim 3, wherein the clock synchronization system is determined according to the number of times.

[5] The correction calculation unit calculates the variation coefficient c when (To−Ti)> 0 continues N times, c =

5. The clock according to claim 4, wherein (previous c) + k is defined, and c = (previous c) —k when (To—Ti) <0 continues N times. Synchronous system.

[6] The clock synchronization according to claim 3, wherein the correction calculation unit defines the variation coefficient c as c = (P set last time) when (To-Ti) = 0. system.

7. The clock synchronization system according to claim 1, wherein the main device transmits the synchronization packet by IP multicast or IP broadcast.

[8] The main device transmits the synchronization packet at regular intervals,

The remote device monitors the estimated arrival time of the synchronization packet received from the main device, and detects the synchronization packet that has arrived beyond a predetermined range from the estimated packet arrival time.

8. The clock synchronization system according to claim 7, wherein the clock synchronization system is excluded from a processing target for controlling a frequency of a clock source capable of variable frequency control.

[9] When the difference (To-Ti) exceeds a predetermined threshold R, the remote device uses the accumulated cumulative time stamp value To and the accumulated remote time stamp value Ti that have been accumulated so far. The clock synchronization system according to claim 7, wherein the clock synchronization system is reset and newly accumulates a time stamp value.

[10] A clock synchronization method for performing clock synchronization between a main apparatus and a remote apparatus that performs voice transmission via a network, using synchronization packets,

The main unit is

Generate a time stamp to generate time stamp information from the clock source,

Sending the synchronization packet in which the time stamp information is embedded to the network;

The remote device is:

The network power receiving the synchronization packet;

The main unit cumulative type stamp value To, which is the cumulative time stamp value of the synchronization start power, is calculated from the time stamp information included in the synchronous packet,

Clock source capable of variable frequency control Extracting time stamp information and starting synchronization The remote device cumulative time stamp value Ti, which is the cumulative time stamp value of the remote device, is calculated, and the frequency of the clock source capable of variable frequency control is calculated as the main device cumulative time stamp value To and the remote device cumulative time stamp value Ti. The clock synchronization method is characterized in that correction is performed according to the difference between the two.

A remote device that is provided in a system that performs voice transmission via a network and that performs clock synchronization using a synchronization packet received from a main device,

A packet receiving unit for receiving the synchronization packet in which the generated time stamp information is embedded, and the network power;

A remote device characterized by comprising: