JP2708453B2 - Audio signal processing device - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio signal processing apparatus capable of reproducing a high-quality audio faithfully to an original audio, and more particularly, to dividing audio data into fixed block lengths, The present invention relates to an audio signal processing device that can be applied to packet communication or the like in which a control signal is added for each transmission of information.

[Conventional technology]

In voice signal communication such as voice packet communication, a method of sending only voiced packets (silent portion compression) is generally used to increase the efficiency of use of a communication channel. However, when the compressed silence is reproduced, if the background noise is not taken into account, the decoded sound will sound as if the line was cut off at the silence, which is extremely unnatural.

Conventionally, as a method for solving this unnaturalness, Japanese Patent Laid-Open No.
As seen in 190999 and the like, there has been proposed a method of reproducing information predetermined on the receiving side as a noise level based on temporal information of a silent section of a header portion of a sound packet.

[Problems to be solved by the invention]

In the above prior art, no consideration is given to the noise level in a silent section during a conversation. That is, the background noise level generated on the receiving side is different from the actual background noise,
Further, the sound level of the background noise in normal conversation generally changes every moment, and the reproduction of a certain level of noise on the receiving side according to the conventional technique may cause a decrease in sound quality.

An object of the present invention is to treat the background noise that changes every moment as one piece of sound information, thereby enabling transmission of the actual background noise level and updating of the background noise, thereby achieving improvement in the quality of reproduced voice. .

[Means for solving the problem]

The configuration of the present invention for achieving the above object has the following features.

1. Each time it is determined that there is no sound when the input audio signal is packetized, one packet or a plurality of packets are sent out as valid packets, and the contents are stored in the receiving memory.

2. The information stored in this memory is repeatedly reproduced until the next voice packet arrives, thereby reproducing a silence portion of an arbitrary time length according to the actual noise level.

[Action]

Since a silent packet having background noise information is transmitted every time it is determined that there is no sound, even if a silent section occurs again,
Background noise can be updated by transmitting new noise information as a packet and rewriting the information in the memory.

〔Example〕

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

FIG. 1 is a conceptual block diagram showing an embodiment for realizing the present invention. In FIG. 1, a transmitting side includes a voice input terminal 11 and a PCM encoder 11 for PCM encoding an input analog signal.
And a packet assembling unit 13 for dividing the PCM signal into a predetermined packet length, a voice / silence determining circuit 14 for determining whether the voice packet is a voice packet or a voiceless packet, and a packet for transmitting the packet to the transmission line 16. It comprises a transmission unit 15. On the other hand, the receiving side determines the presence / absence of a packet in the packet receiving section 16 and the received packet, and the voice / non-speech determining circuit 18 and the determining circuit 18 for controlling the sub memory 19 and the selector 20 to be described later. A sub-memory 19 for storing silence section information to be described later, and output information 106 of the determination circuit 18
And a selector 20 for selecting the output information 108 of the sub-memory 19 and outputting it to the buffer 21 and a buffer 21 for absorbing the in-network delay variation of the voice packet sent from the selector 20.
And a packet decomposing unit 22 for converting a packet into a PCM signal.
It comprises a PCM decoder 23 for converting a CM signal into an output analog signal and an audio output terminal 24.

The analog audio signal 101 input to the audio input terminal 11 is P
It is converted into a PCM signal 102 by the CM encoder 12. PCM signal 102
Are assembled into voice packet information 103 of a predetermined length by the packet assembling section 13. Further, the packet information 103 is subjected to sound / silence judgment by comparing the packet information value with a preset silence judgment value in a sound / silence judgment circuit 14. The packet determined to be sound by the determination circuit 14 is sent to the packet transmission unit 15 as valid packet information 104. On the other hand, following the packet information determined to be sound by the determination circuit 14, the packet information determined to be silent is sent to the packet transmission unit 15 for background noise reproduction. The second and subsequent consecutive packets determined to be silent by the determination circuit 14 are discarded. As described above, all the packets sent to the packet transmission unit 15 are valid packets, and the packet transmission unit 15 transmits these packets to the transmission path.
Send to 16 in order.

The voice packet arriving at the packet receiving unit 17 is sent to the voice / non-voice determination circuit 18. The received voice packet 105 is sent to the comparing circuit 18- (b) in the judging circuit 18 and at the same time, is sent out as an output 106 from the judging circuit 18 to the selector 20 and an output 107 to the sub memory 19. The judgment circuit 18 compares the output 113 of the silence judgment value storage section 18- (a) with the size of the voice packet 105 to make a sound / silence judgment.
When the voice packet 105 has sound, the judgment circuit 18 prohibits the writing of the output 107 to the sub memory 19 by the sub memory control line 115, and the selector 20 controls the output 106 by the selector control line 114. Instruct the user to make a selection.
On the other hand, when the voice packet 105 is silent, the sub memory control line 115 instructs the writing of the output 107 to the sub memory 19, and the selector control line 114 selects the output 108 from the sub memory 19. To instruct. The control by the selector control line 114 is continued until a new packet arrives.For example, when silent portion compression is performed on the transmission side as described above, the control is continued until the next voiced packet arrives at the determination circuit 18. The contents written in the sub-memory 19 are sequentially selected and output 109.

The packet group 110 whose delay has been adjusted by the buffer 21 is converted into a PCM signal 111 by a packet decomposing unit 22, decoded by a PCM decoder 23 into an analog signal, and then output from an audio terminal 24.

FIG. 2 is a time chart schematically showing the behavior of a voice packet on the network. In FIG. 2, in each of the packet groups, the inclined packet indicates a sound packet.
Open packets indicate silence packets. The numbers assigned to the packets indicate the order in which the packets are created.

In FIG. 1, a packet information output 103 generated by the packet assembling section 13 is sent to a sound / non-speech determination circuit 14. The determination circuit 14 determines the presence / absence of sound in the packet information output 103, and the second and subsequent silent packet information 209 and 21 in the silent section.
Discard 0,211. That is, the packet information 104 is assembled and transmitted to the transmission line 16 as a packet.

The packet information 105 arriving at the receiving side is sent to the sound / non-speech determination circuit 18. First packet information 20 which is sound information
The first packet information 202 is sent to the buffer 21 after having been subjected to the sound judgment by the judgment circuit 18. The third packet 203 (a), which is a silence packet, is subjected to silence judgment by the judging circuit 18 and is written into the sub-memory 19 and sent to the buffer 21. The information 203 (b) of the third packet stored in the sub-memory 19 is read out as the information of the fourth packet 204 and the fifth packet 205 according to the instruction of the determination circuit.
If the determination circuit 18 determines that the sixth packet information 206, which is new voiced packet information, is voiced, the packet 206
Is sent to the buffer 21. The seventh packet information 207 (a), which is a new silence packet, is subjected to silence judgment by the judging circuit 18, is newly written in the sub memory 19, and then output to the buffer 21. The seventh packet information 207 (b) stored in the sub memory 19 is stored in the determination circuit 1
The information is read out as the information of the eighth packet 208 according to the instruction of 8.

The packet information 104 compressed by the sound / non-speech determination circuit 17 as described above is output from the buffer 21 as packet information 109 corrected based on the third packet information 203 (a) and the seventh packet information 207 (b). Is output.

Table 1 shows an example of the length of silence in a standard conversation.
This figure shows the effect of the present embodiment (sending one packet having background noise information in a silent section) on silent section compression with respect to the conventional method that does not send silent information.

Here, μ indicates the average value and σ indicates the standard deviation. Assuming that 32 Kbps ADPCM is used for the encoder and 31 bytes are used for the voice data of one packet width, voice for one packet is equivalent to 7.75 msec.

The average silence time μ is as long as 1.7 seconds (equivalent to 219 packets), and the loss according to this embodiment is only 0.46%. Also, even when the variation in data is taken into account, the loss of silent section compression is 0.70% or less. A remarkable increase in the loss of silence compression greatly affects the network utilization efficiency. However, the loss is as small as this embodiment, and the effect is extremely small.

Also, as an application example of the present embodiment, in FIG. 1, after the voice packet information 103 is determined by the transmission side voice / non-voice determination circuit 14, the determination result is output to the header of the output packet 104 from the packet transmission unit 15. (For example, 1 for sound and 0 for silence)
A method is conceivable. In this case, the sound / silence judgment of the reception packet 105 in the sound / silence judging circuit 18 on the receiving side is sufficient to judge by 1 bit, and the silence judgment value on the transmitting side need not be known in advance. The processing in the circuit 18 can be simplified.

In this embodiment, only the first packet of the silent part is transmitted. However, the number of packets to be transmitted is a problem relating to the efficiency of the silent part compression and is not essential for the present invention.

Further, the present embodiment has shown the reproduction of a silent part following a sound part. For example, even when a silent part occurs at the same time as call setting, the silent part is detected and background noise information is sent out to reproduce the silent part. Reproduction is possible.

〔The invention's effect〕

According to the present invention, since a silent packet is transmitted as background noise information, approximate reproduction of actual noise can be performed and noise can be updated, so that there is an effect of improving the quality of reproduced voice.

[Brief description of the drawings]

FIG. 1 is a conceptual block diagram showing an apparatus configuration of an embodiment of the present invention, and FIG. 2 is a time chart schematically showing the behavior of a voice packet on a network. 11 ... voice input terminal, 12 ... PCM encoder, 13 ... packet assembling unit, 14 ... sending / non-speech determining circuit on the transmitting side, 15 ... packet transmitting unit, 16 ... transmission line, 17 ... packet Receiving section, 18: Receiving side sound / non-speech determination circuit, 18 (a): Silence determination value storage section, 18 (b), comparison circuit, 19: Sub memory, 20: selector, 21 buffer, 22 ... bucket decomposition unit, 23 ... PCM decoder, 24 ...
Audio output terminal, 114 ... selector control line, 115 ... sub-memory control line.

Continuation of front page (56) References JP-A-63-124636 (JP, A) JP-A-57-190999 (JP, A) JP-A-63-257367 (JP, A) JP-A-55-33347 (JP, A) , A)

Claims (2)

(57) [Claims] 1. An audio signal processing apparatus comprising: a transmitting side for digitizing an input audio signal, dividing the input audio signal into blocks of a predetermined length, and transmitting the blocks; and a receiving side for receiving the blocks and reproducing the audio signal. The side has voiced / silent determination means for determining voiced / silent of the block, and outputting a voiced block and at least one voiceless block following the voiced block as a valid block. A voice / silence determining means for determining voice / silence, and a memory, wherein when the voice / silence determining means determines that there is voice, the valid block is reproduced, and when it is determined that there is no voice, Wherein the effective block is stored in the memory, and the block stored in the memory is repeatedly reproduced until a next effective block is received. 2. The sound / silence determining means on the transmitting side adds a determination result to the block and transmits the block, and the sound / silence determining means on the receiving side performs processing with reference to the determination result. The audio signal processing device according to claim 1, wherein: