JP3397191B2 - Delay fluctuation absorbing device, delay fluctuation absorbing method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a delay fluctuation absorbing device and a delay fluctuation absorbing method, and in particular, when performing voice communication using a packet communication network typified by the Internet, it absorbs delay fluctuation generated by network traffic or the like. The present invention relates to a voice communication fluctuation absorbing device and a voice communication fluctuation absorbing method.

2. Description of the Related Art In recent years, packet communication networks have become popular with the rise of the Internet. This packet communication network has been developed mainly for data communication. The packet communication network has a feature that the transmission delay becomes large when the communication band is narrow and the transmission delay becomes small when the communication band is wide. Furthermore, when bursty traffic occurs in the packet communication network, the transmission delay increases only at that moment. Therefore, when voice is transmitted as data in the packet communication network, delay fluctuation occurs in the voice to be transmitted due to the above characteristics.

On the other hand, a voice communication network used in a telephone or the like has a feature of transmitting and receiving voice data by a fixed amount at fixed time intervals.

Therefore, a voice reproduction device such as a telephone set or a device for outputting voice data transmitted on the packet communication network to the voice communication network has a function of absorbing delay fluctuations generated in the packet communication network. Is required.

By the way, there are various protocols used for packetizing voice data, and one of the typical protocols is RTP (Real-time Transport Pr).
otocol). This RTP is for adding a sequence number and a packet transmission time to voice data when transmitting the voice data into a packet and transmitting the packet to a packet communication network. Then, the receiving side of the voice packet absorbs the delay fluctuation based on the sequence number and the packet transmission time added to the voice data.

Further, in a device which receives voice data from a packet network and sends it to a voice communication network, a FIFO buffer is generally used in order to absorb delay fluctuation. In order to absorb a large delay fluctuation, the time from the input of the head voice packet to the output of the FIFO buffer to the output thereof is increased, that is, the delay amount in the FIFO buffer is increased. However, an unnecessarily large delay makes the voice exchange unnatural and hinders the conversation in voice communication. Therefore, the delay amount needs to be the minimum amount necessary to absorb the delay fluctuation.

Therefore, as a technique for reducing the delay amount to a necessary minimum amount, there is a technique disclosed in Japanese Patent Laid-Open No. 11-41287 (hereinafter referred to as Prior Art 1). In this prior art 1, when the audio data accumulated in the FIFO buffer exceeds a certain amount, the time exceeding the certain amount continues for a certain time, and when the silence state of the accumulated audio data is detected, the resetting means is provided. Is characterized in that the storage capacity of the FIFO buffer is cleared to zero.

Further, as another conventional technique for absorbing the delay fluctuation generated in the packet communication network, there is a technique disclosed in FIG. 4 (hereinafter referred to as conventional technique 2). The delay fluctuation absorbing device disclosed in the prior art 2 is a FI that temporarily stores voice packets in order to absorb the delay fluctuation.
The FO buffer 20 and the jitter calculation unit 2 that calculates the delay fluctuation amount by comparing the transmission time and the reception time of the voice packet.
1 and the timing of reading a voice packet from the FIFO buffer 20, and a delay adding unit 22 having a function of inserting a silent packet when increasing the delay amount and discarding voice data when decreasing the delay amount.
And a determination unit 23 that gives an instruction to increase or decrease the delay amount added by the delay addition unit 22 based on the delay fluctuation amount calculated by the jitter calculation unit 21.

The voice data received from the packet communication network is accumulated in the FIFO buffer 20, and at the same time, the jitter calculating section 21 compares the transmission time and the reception time to obtain the jitter.
(Amount of delay fluctuation) is calculated. The determination unit 23 obtains a delay amount that is large enough to absorb the delay fluctuation calculated by the jitter calculation unit 21. When the calculated delay amount is larger than the current set value by a certain value or more, the delay adding unit 22 is instructed to increase the delay amount. On the other hand, when the calculated delay amount is smaller than the current set value by a certain value or more, the delay adding unit 22 is instructed to reduce the delay amount. Based on the instruction from the determination unit 23, the delay addition unit 22 adjusts the delay amount by inserting a silent packet or discarding a voice packet. In some cases, noise packets may be inserted instead of silent packets, or one voice packet may be repeatedly transmitted.

However, the prior arts 1 and 2 have the following problems.

First, in the prior art 1, since the buffer storage amount is fixed, it is necessary to set the storage amount large. Therefore, there is a problem that the delay amount becomes large and the voice delay becomes large, which hinders a natural conversation. Further, since the buffer storage amount is fixed, there is also a problem that it is not possible to flexibly cope with fluctuations in delay fluctuation of the packet communication network.

In the prior art 2, when the delay adding unit 22 increases the delay, pseudo packets, that is, artificially generated silence packets and noise packets are repeatedly inserted, and therefore the transmitting side does not transmit the packets. Since the voice data is transmitted, there is a problem that the voice data reaching the telephone set or the voice communication network is mixed with a sound that does not originally exist, which causes a feeling of strangeness. Therefore, the present invention has been invented in view of the above problems, and an object of the present invention is to provide a delay fluctuation absorbing device capable of increasing / decreasing the delay addition amount without impairing the clarity of voice. And a method for absorbing delayed fluctuations.

SUMMARY OF THE INVENTION An object of the present invention is a delay fluctuation absorbing device for absorbing delay fluctuations of voice data sent from a packet communication network. A buffer that temporarily stores, a delay fluctuation calculation unit that measures the amount of delay fluctuation of the arrived voice packet, and a voice packet stored in the buffer is extracted, and silence is detected for the retrieved voice packet, A silence detecting means for adding silence information indicating silence to a silent voice packet, a delay fluctuation amount measured by the delay fluctuation calculating means, and a set delay setting value are compared, and the measured delay is compared. When the fluctuation amount is larger than the delay setting value by a predetermined value or more, an instruction to increase the delay amount is issued, and the measured delay fluctuation amount is set to the delay setting value. Is smaller than the predetermined value by a delay amount control means for issuing an instruction to decrease the delay amount, and a voice to which the silent information is added when receiving the instruction from the delay amount control means and increasing the delay amount. Duplicate the packet
And a delay amount adjusting means for adjusting the delay amount by discarding the voice packet to which the silent information is added when reducing the delay amount. It

According to the present invention, the voice detecting means detects a silent voice packet from the voice packet, and adds silence information indicating the silence to the detected voice packet. Then, the delay amount control means compares the delay fluctuation amount measured by the delay amount calculation stage with the set delay setting value, and the measured delay fluctuation amount is a preset value rather than the delay setting value. If it is larger than the above, an instruction to increase the delay amount is issued, and if the measured delay fluctuation amount is smaller than the delay setting value by a preset value or more, an instruction to decrease the delay amount is issued. Based on this, the delay amount adjusting means repeatedly transmits the voice packet to which the silent information is added when increasing the delay amount, and discards the voice packet to which the silent information is added when decreasing the delay amount. It is characterized by adjusting the delay amount.

It is to be noted that the delay fluctuation absorbing device extracts the voice packet accumulated in the buffer, detects a voice packet having no change in pitch, volume or sound quality with respect to the retrieved voice packet, and detects the detected voice packet. A unit for adding non-change information indicating that there is no change in pitch, volume, or sound quality to the packet, and the non-change information is added when the delay amount is increased in response to an instruction from the delay amount control unit. It is also possible to further provide a second delay amount adjusting means for adjusting the delay amount by discarding the voice packet to which the unchanged information is added when repeatedly transmitting the voice packet and reducing the delay amount.

The delay amount control means compares the delay fluctuation amount measured by the delay fluctuation calculation means with a set delay setting value, and the measured delay fluctuation amount is smaller than the delay setting value. Is greater than a preset second predetermined value, an instruction to repeatedly send a voice packet to which silence information is not added is also issued, and the delay fluctuation amount is set to a preset preset value larger than the delay set value. If it is smaller than the predetermined value of 2, the voice packet to which the silent information is not added is instructed to be discarded, and the delay amount adjusting means is a voice packet to which the silent information is not added. When receiving an instruction to repeatedly send the voice packet, the voice packet to which the silence information is not added is also repeatedly transmitted, and the voice packet to which the silence information is not added is also discarded. When receiving the instruction that may also be configured to discard even voice packets silence information is not added.

Further, the delay fluctuation calculation means can be configured to calculate the delay fluctuation amount based on the transmission time added to the voice packet and the reception time when the voice packet is received. is there.

Voice packets are RTP (Real-time).
It is preferably packetized based on Transport Protocol).

The above object of the present invention is a delay fluctuation absorbing method for absorbing delay fluctuations of voice data sent from a packet communication network, in which voice packets sent from the packet communication network are temporarily stored. And a step of measuring the delay fluctuation amount of the arrived voice packet, extracting the accumulated voice packet, detecting silence in the retrieved voice packet, and indicating silence in the silent voice packet. The step of adding
The measured delay fluctuation amount is compared with a set delay setting value, and if the measured delay fluctuation amount is larger than the delay setting value by a predetermined value or more, the voice to which the silent information is added is added. A step of adjusting the delay amount by duplicating and transmitting the packet, and discarding the voice packet to which the silent information is added when the measured delay fluctuation amount is smaller than the delay setting value by a predetermined value or more. Is achieved by the delayed fluctuation absorption method.

In the delay fluctuation absorbing method, the accumulated voice packets are taken out, voice packets having no change in pitch, volume or sound quality are detected from the taken voice packets, and the detected voice packets are detected. A step of adding unchanged information indicating that there is no change in pitch, volume, or sound quality, and when increasing the delay amount, repeatedly transmitting the voice packet to which the unchanged information is added, and decreasing the delay amount May further include the step of discarding the voice packet to which the unchanged information is added.

In the delay fluctuation absorbing method, when the measured delay fluctuation amount is larger than the delay set value by the second predetermined value or more, the voice packet to which the silent information is not added is repeatedly transmitted. However, when the measured delay fluctuation amount is smaller than the delay setting value by a second predetermined value or more, it is possible to further include a step of discarding a voice packet to which silence information is not added.

The step of measuring the delay fluctuation amount may be a step of calculating the delay fluctuation amount based on a transmission time added to the voice packet and a reception time when the voice packet is received. preferable.

The voice packet is RTP (Real-time).
It is preferably packetized based on Transport Protocol).

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described.

The voice data transmitted on the packet communication network is generally packetized using an image and voice communication protocol typified by RTP (Real-time Transport Protocol). Also in this embodiment, the audio data transmitted on the packet network will be described as being packetized using RTP.

FIG. 1 is a block diagram showing this embodiment.

In FIG. 1, reference numeral 1 is a buffer for temporarily accumulating a voice packet a input from a packet communication network, and a FIFO (First-in-First-Out) type buffer is used.

Reference numeral 2 denotes a jitter calculator, which calculates a jitter b (delay fluctuation amount b) of the arrived voice packet a as the voice packet a arrives. As a calculation method, the delay fluctuation amount b is calculated by comparing the transmission time added to the reception data with the reception time at which the data is received.

Reference numeral 3 denotes a judging unit, which judges the delay amount c based on the delay fluctuation amount b calculated by the jitter calculating unit.
Is calculated and the instruction d is given to the delay adding unit 5 described later. When calculating the delay amount, a delay amount that is large enough to absorb the calculated delay fluctuation amount b is obtained. As a method of instructing the delay adding unit 5, the calculated delay amount c and the current delay setting value e set in the delay adding unit 5 are compared. Then, when the calculated delay amount c is larger than the delay setting value d by a predetermined value x (set in advance) or more, the delay adding unit 5 is instructed to increase the delay amount. On the other hand, when the calculated delay amount c is smaller than the delay set value e by the predetermined value x or more, the delay adding unit 5 is instructed to reduce the delay amount.

Reference numeral 4 is a silence detector. This silence detector 4
When a voice packet read request f is issued from the delay adding unit 5, the packet is fetched from the FIFO buffer 1 (request g), and it is detected whether or not the fetched packet is silent. When the extracted packet h is silent, a voice packet h is flagged as silence as silence information indicating that the packet is a silence packet.

Reference numeral 5 is a delay adding unit. This delay addition unit 5
Repeats the silent packet based on the instruction d of the determination unit 3,
Alternatively, the amount of delay is increased or decreased by discarding. That is, when the received instruction d is an instruction to increase the delay amount,
A voice packet (silent packet) to which a flag indicating silence is added and one copy thereof are transmitted. On the other hand, if the received instruction d is an instruction to reduce the delay amount, one voice packet to which a flag indicating silence is added is discarded.

Next, the operation of the present embodiment configured as described above will be described with reference to the flowchart of FIG. In the present description, the above-mentioned predetermined value x is 1
It will be described as 5 msec.

First, the voice packet a received from the packet communication network is stored in the FIFO buffer 1 (Step 100), and at the same time, the jitter calculating unit 2 calculates the delay fluctuation amount b (Step 101).

The determination unit 3 obtains a delay amount large enough to absorb the calculated delay fluctuation b, and the calculated delay amount c and the current delay set value d set in the delay addition unit 5.
And are compared (Step 102). When the calculated delay amount c is larger than the delay setting value e by 15 msec or more, an instruction d is issued to the delay adding unit 5 to increase the delay amount (Ste).
p 103). The calculated delay amount c is 15 than the delay setting value e.
If it is smaller than msec, an instruction d is issued to the delay adding unit 5 to reduce the delay amount (Step 104).

When there is a voice packet read request f from the delay adding unit 5, the silence detecting unit 4 takes out a voice packet from the FIFO buffer 1 (Step 105).
It is detected whether the extracted voice packet is silent (Step 106). When it is determined that the voice packet is silent, a flag indicating that is set in the voice packet and sent to the delay adding unit 5 (Step 10).
7). On the other hand, when it is determined that the voice packet is not silent, the voice packet is sent to the delay adding unit 5 without setting a flag (Step 108).

The delay adding section 5 increases or decreases the delay amount based on the instruction from the judging section 3. If the delay is to be increased, the voice packet with the silence flag and one copy thereof are transmitted (Step 109). When reducing the delay, one voice packet with a silence flag is discarded (Step
p 110).

By operating in this manner, when the present invention increases the delay addition amount, the transmitting side repeatedly reproduces the silent portion of the transmitted voice packet, so that the silent portion between the voices is reproduced. Since the delay amount is increased by playing for a long time, it is possible to maintain the clarity of the conversation without making the reproduced sound unnatural compared to the method of inserting artificially created noise or the method of randomly repeating the voice. You can

In the description of this embodiment, the predetermined value serving as the reference of the instruction d for increasing / decreasing the delay amount has been described as 15 msec, but this is merely an example, and when the voice data is actually packetized. The value varies depending on the packetization time interval and the transmission characteristics of the packet communication network.

Further, in the description of the present embodiment, the repeating voice packet (silent packet) is described as one, but two or more duplicates may be duplicated and repeatedly transmitted in consideration of the system state. Further, in the description of this embodiment, two or more packets to be discarded may be discarded in consideration of the system state.

Further, when the determination unit 3 issues an instruction d for increasing or decreasing the delay amount to the delay adding unit 5, if the calculated delay amount c and the delay setting value e are significantly different, for example, in the above example, 30 msec (second When there is a difference equal to or more than a predetermined value), not only the voice packet with the silence flag set but also the normal voice packet may be discarded or repeated. In this case, when the difference between the calculated delay amount c and the delay setting value e is equal to or more than the second predetermined value (for example, 30 msec), the determination unit 3 determines that not only the voice packet with the silence flag is set. It is configured to issue an instruction to discard or repeat even a normal voice packet. Then, the delay adding unit 5 that has received the instruction discards not only the voice packet with the silence flag set but also the normal voice packet,
Alternatively, it may be configured to execute the repetition. It should be noted that the reference value (first predetermined value) of the difference between the calculated delay amount c and the delay setting value e for starting the discarding of only the voice packet with the silent flag set, and the calculation delay for starting the discarding of the normal voice packet The reference value (second predetermined value) of the difference between the amount and the delay setting value is
They may be set the same or may be set independently.

With this configuration, it becomes possible to absorb delay fluctuations even for voices with few silent sections.

Next, another embodiment will be described.

FIG. 3 is a block diagram of another embodiment of the present invention.

The difference between the other embodiments and the above-mentioned embodiment is that a long sound detecting section 10 is further added as shown in FIG.

When the long sound detecting section 10 detects a voice packet having no change in pitch, volume, or sound quality, it flags the voice packet. The delay adding unit 5 handles the voice packet in the same manner as a voice packet with a silence flag.

With such a configuration, a portion having a constant pitch, volume, and sound quality is not important in general conversation, and the clarity of the conversation is not impaired.

[0043]

According to the voice communication fluctuation absorbing method of the present invention,
In order to absorb voice communication fluctuations, the silent portion is repeatedly reproduced when increasing the delay addition amount. The delay amount is increased by playing the silent part between voices for a long time instead of the voice during conversation.Therefore, it is possible to play it by a method of inserting artificially created noise or a method of randomly repeating the voice. The sound is not unnatural and the clarity of conversation can be maintained.

When the delay addition amount is reduced, the silent portion is discarded, so that the clarity of the conversation is not impaired in the same manner.

[Brief description of drawings]

FIG. 1 is a block diagram of the present embodiment.

FIG. 2 is an operation flowchart of the present embodiment.

FIG. 3 is a block diagram of another embodiment.

FIG. 4 is a diagram for explaining a conventional technique.

[Explanation of symbols]

1 FIFO buffer 2 Jitter calculator 3 Judgment section 4 silence detector 5 Delay addition section

Claims (10)

(57) [Claims] 1. A delay fluctuation absorber for absorbing delay fluctuations of voice data sent from a packet communication network, wherein a buffer for temporarily storing voice packets sent from the packet communication network arrives. Delay fluctuation calculating means for measuring the amount of delay fluctuation of the voice packet, extracting the voice packet accumulated in the buffer, detecting silence in the retrieved voice packet, and silent information indicating silence in the silent voice packet. A silence detecting means for adding, a delay fluctuation amount measured by the delay fluctuation calculating means,
If the measured delay fluctuation amount is larger than the delay setting value by a predetermined value or more, an instruction to increase the delay amount is issued, and the measured delay fluctuation amount is compared with the set delay setting value. Is smaller than the delay set value by a predetermined value or more, a delay amount control unit that gives an instruction to decrease the delay amount; and a silent amount information when receiving an instruction from the delay amount control unit and increasing the delay amount. Voice packet with
A delay fluctuation absorbing device comprising: a delay amount adjusting unit that adjusts the delay amount by discarding the voice packet to which the silent information is added when the data is copied and transmitted and the delay amount is reduced. 2. A voice packet stored in the buffer is taken out, and a voice packet having no change in pitch, volume or sound quality is detected from the taken voice packet,
A unit for adding unchanged information indicating that there is no change in pitch, volume, or sound quality to the detected voice packet, and the unchanged information when the delay amount is increased in response to an instruction from the delay amount control unit. And a second delay amount adjusting means for adjusting the delay amount by repeatedly transmitting the voice packet added with and discarding the voice packet added with the unchanged information when reducing the delay amount. The delay fluctuation absorbing device according to claim 1, wherein: 3. The delay amount control means compares the delay fluctuation amount measured by the delay fluctuation calculation means with a set delay setting value, and the measured delay fluctuation amount is smaller than the delay setting value. Is greater than a preset second predetermined value, an instruction to repeatedly send a voice packet to which silence information is not added is also issued, and the delay fluctuation amount is set to a preset preset value larger than the delay set value. If it is smaller than the predetermined value of 2, the voice packet to which the silent information is not added is instructed to be discarded, and the delay amount adjusting means is a voice packet to which the silent information is not added. When receiving an instruction to repeatedly transmit the voice packet, the voice packet to which the silence information is not added is also repeatedly transmitted, and the voice packet to which the silence information is not added is also discarded. When subjected to indicate the delay fluctuation absorbing device according to claim 1 or claim 2, characterized in that voice packets silence information is not added is also configured to discard. 4. The delay fluctuation calculating means is configured to calculate a delay fluctuation amount based on a transmission time added to a voice packet and a reception time when the voice packet is received. The delay fluctuation absorbing device according to any one of claims 1 to 3, which is characterized. 5. A voice packet is RTP (Real-time Tracing).
5. The delay fluctuation absorbing device according to claim 1, wherein the delay fluctuation absorbing device is packetized based on the nsport protocol). 6. A delay fluctuation absorbing method for absorbing delay fluctuations of voice data sent from a packet communication network, which comprises temporarily storing voice packets sent from the packet communication network, and arrived. Measuring a delay fluctuation amount of the voice packet; extracting the accumulated voice packet, detecting silence in the retrieved voice packet, and adding silence information indicating silence to the silent voice packet; The measured delay fluctuation amount is compared with a set delay setting value, and if the measured delay fluctuation amount is larger than the delay setting value by a predetermined value or more, the silence information is added. sending duplicates the voice packet, the silence information added when the measured delay fluctuation amount is smaller than a predetermined value than the delay setting value Delay variation absorbing method characterized by a step of adjusting the delay amount by discarding voice packets. 7. The stored voice packet is extracted, a voice packet having no change in pitch, volume or sound quality is detected from the extracted voice packet, and the detected voice packet is changed in pitch, volume or sound quality. There is a step of adding unchanged information indicating that there is no change, and when the delay amount is increased, the voice packet to which the unchanged information is added is repeatedly transmitted, and when decreasing the delay amount, the unchanged information is added. The method according to claim 6, further comprising: discarding the added voice packet. 8. If the measured delay fluctuation amount is larger than the delay setting value by a second predetermined value or more, a voice packet to which silence information is not added is repeatedly transmitted, and the measured delay 8. The method according to claim 6, further comprising the step of discarding a voice packet to which silence information is not added when the fluctuation amount is smaller than the delay setting value by a second predetermined value or more. Delay fluctuation absorption method. 9. The step of measuring the delay fluctuation amount is a step of calculating the delay fluctuation amount based on a transmission time added to the voice packet and a reception time when the voice packet is received. The delay fluctuation absorbing method according to any one of claims 6 to 8, which is characterized. 10. A voice packet is RTP (Real-time T).
10. The delay fluctuation absorbing method according to claim 6, wherein the method is packetized based on the ransport protocol).