JPS5850597A - Voice storage system - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a voice storage system in exchanges that provide voice mail services, voice output services, etc. It provided direct communication between the caller and the called party. However, in Kumi's situation, the person she wants to talk to may be absent, or may be talking on another phone, or the other person's phone may be busy, so the task cannot be completed in one go.

This problem is particularly important in business communications, with research reporting that the probability of successfully communicating with a desired party is 30%, 8 times. KFJ has installed a voice storage device in the exchange, and when it is necessary to call the other party up to the telephone and speak directly, there are many cases where it is not necessary to call the other party to the telephone and speak directly. If, also i[
If the message does not require a close conversation, the message will be stored as voice and the exchange will deliver the message to the other party on behalf of the caller at a desired time, or at a time convenient for the other party. It is hoped that the meeting will provide services that will attract people to come and listen to the Metsuryuji. This is called voice mail-1-service, and is considered to be one of the important service functions that future exchanges should focus on.

Conventionally, tone signals such as key, dial tone, and busy tone have been used to inform callers of call progress and the like from exchanges. Therefore, it is thought that the types of information that would be useful to be notified from the exchange to the caller will rapidly increase, and it will be extremely difficult to identify them using tone signals alone. Therefore, switching equipment needs to have the ability to output this information in the form of voice. Although it would be most ideal to use a voice synthesizer for K to realize the voice output function, the voice quality etc. are still not sufficient at present. The most practical method is to store the audio information to be output in advance in the exchange and read it repeatedly as needed.

As mentioned above, whether it is a voice mail service or a voice output function, it is required to have an in-exchange Kf voice storage system. method and storage capacity. As a storage method, it is conceivable to use the ANAMIG method to record the sound on a tape recorder, but it is inappropriate to have two random outputs and repeated writes. - It is advantageous to treat data and files in the same way as data and files are processed in a computer. Given that digital exchanges will become mainstream in the future, it will be even more advantageous to handle digitized voice.However, in that case, storage capacity will be an issue. According to the world's IIA quasi-method (PCM method) for audio signal digitization, the audio signal is 64kb per second.
In other words, it has an amount of information of 8 bytes.To store one minute's worth of voice messages, a storage capacity of 480 bytes, or approximately 0.5M bytes, is required.A very large number of messages are stored in the entire exchange. Considering this, it is necessary to take measures to somehow reduce this storage capacity.

One of the countermeasures is to change the digitization method from the PCM method.
A set of modulation methods that require a small amount of information, such as Amebitibu Delta Modulation (DM) method, adaptive differential PCM (A
This is a method of performing so-called band compression, converting to a DPCM (DPCM) system, etc., and then storing the data. Even now, very good sound quality can be obtained at the PCM speed of 32kb/sO, and it is expected that 16kb/sO will become available in the future.

Another measure to reduce the storage capacity is to remove and store the silent periods that exist between human speech. It is thought that by removing the silent sections that exist when thinking, it is possible to save the flooring capacity by one or more units.

However, let's talk about silent sections like this! When a WAK is stored as a sound section, a method for reproducing the silent section is required for the WAK that plays it. the position and length of
It is necessary to memorize and manage data such as boundary points of sound sections on the storage device.

Figure 1 shows a conventional example of the audio storage method in which the KIIA sound section is removed, stored, and played back as described above.
(Be1l 8ystsm Technical
Journal.

0ctober 1980 p, 1383-1395
'An & parimenta18p@ech Sto
rag@and FJiting Facility
@Reference) In the audio storage method shown in No. 111, the digitized audio code applied to the input terminal 10 is
Sampling the audio code applied to the input terminal 11 DM the clock signal with the frequency
A (Direct Memory Ace@ss) clock is a buffer memory 13 provided on the main memory of the converter 12 or t! In 14K" DMA mode, switching between the buffer memory 13 and Fi 14, where 0 is written to the logic register) L, is performed by the switch 16 based on the command from the controller 15 of the converter 120.
It is carried out by However, the switch 16 is a logical switch. If the switch 16 is now in the state shown in FIG. 1 and the voice code has been written to the end in the buffer memory 13, a - write is generated to the control section 15 to notify the control section 15 of this fact. Control 1115 responds to this V@include by 2
The de-16 buffer and l memory are set to 14-fold, and the voice code writing tube continues. Meanwhile, the control unit 15Fi performs various calculations based on the voice code sequences written in the buffer memory 13 and determines whether these voice code sequences correspond to a sound section or a silent section. As a result, if it is determined that it is a sound section, the control s15 writes the f′L running audio code to the buffer memory 13 and transfers it to the magnetic disk 17. At that time, switch 18 is buffer memory 1B@
It is in @.

The switch 18 is also a logical switch, and operates in conjunction with the switch 16 to select different buffer memories. On the other hand, if the audio code in the buffer memory 1340 falls into a silent section, If it is determined, propagation from the buffer memory to the magnetic disk is not performed. However, the buffer
When the memory 14 is finally written, writing to the buffer memory 13 is started, so that the previous speech code is erased. Buffer memo while writing buffer memory 13! The operation of the control unit 15 for j14 is also completely similar.

In other words, in the conventional method shown in FIG. 1, the entire voice code written in the buffer memo IJK is treated as one block, and it is determined whether or not it is a sound section in block units, and the block determined to be a sound section ( (sound block) to a magnetic disk. -! The buffer memory size Fi corresponding to one sector of the magnetic disk or an integral multiple thereof is selected in relation to the input/output unit of the magnetic disk.

In addition, in order to correctly insert the silent block into the position of the sound block during playback, the control unit 15 transfers the sound block to the magnetic disk 17 (for example, the serial number from the previous message of the block and the block). Address on the magnetic disk where it was stored (sector number, etc.)'ft
Remember. The information Fi control data is stored in the main memory or in a portion of the magnetic disk 17 that is different from the audio code.

Now, the stored audio is played back as follows.

That is, the control unit 15tj reads the number of the sound block and its address from the control data stored in the main memory or the magnetic disk, takes it out from the sound block magnetic disk 17, and stores it in the buffer memory 13 or #. 'i1
Enter 4. KF if the sound block number is skipped
i. Since there was a silent block during that time, before extracting the sound block from the magnetic disk 17, clear the buffer memory 13 or 14 and insert a predetermined number of silent blocks. do. After that, the sound block tube is taken out and the same operation continues. 0 buffer memory 13.14 ii.When one is read out to the input/output terminal 10 in DMA mode by the switch 16.18, the other is read out to the control section 150. The control allows switching between inputting a sound block from the magnetic disk 17 and clearing a silent block for manual input, as in the case of audio storage. .

Note that the control data includes a method using the block number and address for the spinning speed sound block.
,Yes! Various methods can be considered, such as arranging the L address for a block and the number of consecutive blocks for a silent block. Many parts are determined by hardware factors. That is, since input/output is normally performed in sectors on magnetic disks, the size of the buffer memory must be the same as the sector or an integral multiple thereof. Also, magnetic disk access time, data transfer time,
If the control unit does not have the necessary σ buffer e memory for the audio code in the buffer memory, it will not appear.

Brock Knize Inc. should be determined based on the nature of the audio value I), but as mentioned above, it should not be influenced by the No. Conceivable.

Changes in block size require a reconsideration of the entire system, which makes it difficult to quickly incorporate technological advances and is therefore unsuitable.

Note that method 1 of making the block size independent of the buffer memory size is not possible, but on the contrary, it complicates the controller processing and is not practical.Also, as mentioned earlier, one control unit is In addition to storing and retrieving the code itself, it is necessary to store information regarding the % voiced section and silent section, and manage the storage and playback of audio based on this information01
Perform these detailed table operations for each message,
In addition, if information regarding the sender, destination, and transmission time of the message as a whole is also managed, the operation of the control section becomes significantly complicated, and various restrictions are placed on the individual OII functions.

The present invention eliminates such drawbacks in the prior art and adds a circuit between the buffer memory and the input terminal, thereby making it possible to set the block Q-nize independently of the buffer memory, magnetic disk, etc. , the information regarding silent blocks is inserted into the 1*voice code to obtain a voice storage method that does not require the control section of the conbeater to be involved at all.

That is, according to the present invention, the voice blocks in the switch are divided into blocks, each block is determined as to whether it is a silent block or a voice block, and only the voice block is marked with a flag at the beginning of the block and the time position of the block or its position. Information regarding the number of silent blocks preceding the block is added and stored in the storage device, and during playback, a flag is detected from the code sequence taken out from the storage device to find W4 after the sound block, and the 7 lag A sound storage method is obtained in which a silent block is inserted based on the information K1IF5 regarding the time position or the number of preceding silent blocks associated with the sound block i, and then the sound block i is retrieved from the storage device and reproduced.

The present invention will be explained in detail below with reference to the drawings.

Are Figures 2 and 113 an implementation of the present invention? FIG. 2 shows a part related to audio storage, and -gmFi shows a part related to audio reproduction.

In FIG. 2, the computer-fi 12, buffer memories 13 and 14, and magnetic disks 17 are the same as in FIG. Terminal 1 of the turtle obtained by
The audio code added to 0 is stored in the buffer memory 21KII.
It is stored in the I order. The size of this buffer memory 21 may be set to a value based on the characteristics of the voice code, and this becomes the unit of play.

However, buffer fist memory 21 ri buffer memory 1
3.14, which is divided into two parts, is shown as one for simplicity. On the other hand, the voice detection circuit 22 checks the level of the voice code, the number of polarity inversions, etc., and determines whether the voice block in the buffer memory 21 is on the conductor layer 4.
It is determined whether the block is a sound block or a silent block, and the resulting sound is output at the end of the block. If it is a silent block, the silent block counter 23 is incremented by 1, and the buffer memory 1 in the DMA clock control circuit 24tj combination island 12 is incremented by 1.
Stop sending the DMA clock to 3 and 14, and even if the buffer message 21 is read, it will not be written to the buffer memory in the combinator, and no silent play will occur. It will not be stored.〇On the other hand, if the voice detection circuit determines that it is a sound block, based on its output,
The DMA clock control circuit 24F1 sends three DMA clocks into the conbeater 12 during one sampling period of the voice code at the beginning of the next block. If the DMA clock control circuit 24 is 8 bits, ////////
Output.

This becomes, so to speak, a flag that zeroes out the sound blocks 0* and *.Next, for the second DMA clock, the silent block -
Counter 230 count value tl - Select and output 0 This represents the number of silent clocks that existed in the leap between the sound block and the previous sound block. ◇After this, the silent block counter 23Fi is reset. . Maximum %lK
13t) For the DMA clock, select the audio code at the beginning of the output from the buffer 21 and output it. From now on, the DMA clock control circuit 24FX
The program sends out one DMA stream during the sampling period and sends out four DMA streams during the sample period.

The output of the far memory 21 is transferred to the in-door bar.

The data is written to the file memory 13 or 14.

Buffer memory 13.14 According to the input DMA clock, reads the voice code or ti7 lag, silent program, number, etc. without being aware of its contents. When it becomes full, the control unit 150 controls 1IK) , the stored contents are stored on the magnetic disk 17 and transferred.

In this way, silent blocks are not stored, and sound blocks are stored on the magnetic disk 17 with the K7 lag and information regarding the silent blocks added.

Note that if there is an identification pattern that is the same as the identification pattern used as 7u or da in the audio code, an error will occur in the detection of the beginning of a sound block and the extraction of information for silent block Kll during audio playback. - To prevent this, the fixed pattern detection circuit 26 monitors the voice code, and when the same pattern as the fixed pattern appears, K sends two DMA clocks to the DMA clock control circuit with respect to the It sampling period. Then, the selection circuit adds another fixed pattern and outputs it. Silent block KII added after 72g! If the information that matches the fixed pattern is set to be prohibited, it is possible to distinguish between .72 and a code that matches the fixed pattern in the voice code using the method described above.

FIG. 4 shows the data structure of one sound block as it is written to the buffer memory 3.14 or the magnetic disk 17. However, n is the number of speech codes of block sushi.

The DMA cloak control circuit 24 frequency-divides the clock applied to the input terminal 27 to create a required DMA clock. In FIG. 3, the flag detection circuit 30 is connected to the buffer memory 13 of the computer 2.
T again! Detects a flag from the code 0 output from 14. When a 0, that is, a fixed pattern used for the flag, is detected, another code
An extra KDMA clock is output to generate the next code 1r. If it continues to match the fixed pattern, it is determined that the previous fixed pattern is a flag and that this code extracted next is information about the number of silent blocks Km. This is input to the block counter 31 and also to the selection circuit 3.
2 selects and outputs the noise generation circuit 33 and inserts a silent block. The block counter 31 is a subtraction counter, and the above state continues until this counter becomes OK. During this time, no clock is output from the DMA cloak control circuit 24, and no code is output from the buffers 13 and 14. Note that the noise generation circuit 33 is a circuit that generates a low-level noise signal corresponding to background noise in order to eliminate the unnaturalness of silent blocks. In addition, the gate 34 is fixed) (If information regarding turns and silent blocks appears, it is used to output it to the terminal 10 and display it.) When the block counter 31 reaches O, the DMA clock control is started again. The circuit 24 outputs a clock and the retrieval of the active block from the buffer memories 13, 14 is performed.
The selection circuit 32 selects the audio codes of these sound blocks that are inputted to the voice block 34 and outputs them to the terminal 10.

Note that when the flag detection circuit 30 detects two fixed patterns in succession, as mentioned above, it is not a flag but a voice code. Therefore, it is output as a voice code to the terminal 10 as it is, and from then on By repeating the above operations, it is possible to reproduce the entire audio while correctly inserting silent blocks.

In the above explanation, information about silent blocks is as follows:
Although we have described a method of storing the number of silent blocks preceding a sound block, it is also possible to store the time position of the sound block, that is, the serial number of the block from the beginning of the message. The difference is whether the serial numbers of the blocks are sent or the difference between them is sent.

As can be seen from the above, according to the original tllK,
An audio professional that is the unit that determines whether there is sound or no sound.

The block size can be determined completely independently from the characteristics of the hardware such as the magnetic disk, internal buffer memory, etc. Therefore, the optimal block size can be set based on the characteristics of the audio. Bandwidth compression can be introduced smoothly.

In addition, information about markers indicating the beginning of blocks and silent blocks is embedded in the audio code, and the control unit of the converter and magnetic disk only treat them as tp/ data; There is no need to worry about identifying sound or silence. Therefore, the management of voice compression/expansion and the management of the entire message can be clearly separated, and a flexible and simple system structure can be constructed, so the effect Fi is extremely large. Also, the added circuits are very simple except for the voice detection circuit. The audio detection circuit also uses gage digital signal processing technology.
Advances in LSI technology have made it relatively easy to achieve.
Above all, it is much easier than having the control unit in the storage device do it.
378-69rDSI Echo Sensor) As described above, according to the present invention, rather than adding a small-scale circuit,

[Brief explanation of the drawing]

1g1 is a block diagram showing a conventional audio storage method, FIGS. 2 and 3 are block diagrams showing an embodiment of the present invention, and FIG. 13, 14% 21 Fi buffer memory % IJ, 17fl magnetic disk, 22 Ii audio detection circuit, 24 FiD clock control circuit, 25.32
23.31 is a block counter, and 30 is a flag detection circuit. Susumu Uchihara, Patent Attorney

Claims (1)

[Claims] In the voice storage method in an exchange, during storage, the code sequence 't of the digitized voice signal is divided into blocks each consisting of a fixed number of codes, and each block is determined whether it is a silent block or a voiced block. Only sound blocks,
Beyond the block! 1 with a flag and information regarding the time position of that block or the number of silent blocks preceding that block, and is stored in the storage device, and during playback,
7 lags are detected from the code sequence retrieved from the storage device to find the beginning of the sound block, and a silent block is inserted based on the time position associated with the flag or information on the number of preceding silent blocks Kli. The sound storage method is characterized in that: 1, the sound block tube is then taken out from the storage device and played back;