JPS6177439A - Dsi lockout control system - Google Patents

Abstract

PURPOSE:To attain fine control optimizing sound quality by using a table using a preset lockout generating rate, a voice detection level and a hangover time as parameters and applying lockout control to the voice detection section next time. CONSTITUTION:A voice detector 1 receives a preset voice detection level 50 and a preset hangover time 60 from a voice detection level register RG 5 and a hangover time RG 6, decides the start point/end point of the voice time and the length of the end of speech of the voice in an input M-channel CH digital voice signal 10 to output a voice signal 11. A voice CH counter 2 counts the CH number (t) being the voice of the signal 11, and when t>N, only the NCH is assigned to a transmission CH assigning circuit 3 and outputted. In this case, the counter 2 counts t-N, what CH is locked out is inputted to a lockout genera tion rate generating rate counter 4. A lockout rate A from the counter 4 and parameters of outputs B, C from RG 5, 6 are used as an address, a signal read from a lockout control table ROM 7 is given to the RG 5, 6 and used for the next voice detection.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a DSI (Digital Speech Interboration) device that transmits signals of only active channels in a telephone line, etc. This invention relates to a freeze-out (freeze-out) control method for locking out which input channels and how to lock them out when the number of input channels becomes larger than the number of output channels.

[Prior art]

In a telephone line, DSI (Digital Speech Interpolation), which transmits signals only on channels that are active, has an M channel on the input line side and an M channel on the output line side if you look at it from a switchboard perspective as shown in Figure 3. 3 normal with N channel line.

M is larger than N, specifically M', 2N.

In this case, if people are talking at the same time, the sounds of half the people will not be transmitted, so statistically speaking, the time when one person is talking and the time when they are not talking is about half and half. You can also say. The reason for this is that when a person on the calling side speaks, the person on the receiving end usually only listens, and his or her own voice is transmitted through the transmission path for only half of the time. Therefore, during this time, other people's voices can be transmitted to the transmission path. For example, in FIG. 3, DSI performs control by utilizing the fact that if input channel B can be input as soon as input channel A ends, the overall transmission capacity will be doubled.

In this Sr device, the number of active input ranks where voice is detected, that is, the number of channels is 1 (1 is M
(smaller than) exceeds the number of output channels N, the voices of the excess channels, that is, (t-N) channels, are not transmitted, and the call is interrupted at that moment. This is called a freeze-out. When this shut-out occurs frequently, it becomes very difficult to have a conversation. Therefore, in order to effectively utilize the transmission line by increasing the gain of the DSI, the lockout occurrence rate is usually allowed to be around 0.5%.

It is known that if this is done, the number M of input/drink ranks will statistically become approximately M'', 2N, relative to the number N of output channels.

In such an SI device, if the detection level is increased and only sounds higher than that level are sent through the transmission path, the transmission capacity will be increased, but the sound quality will be degraded because the subtle parts of the voice will be cut off. The detection level for the end of a word is set lower than the detection level for the beginning of a word, so that the two-word final sounds are delayed for a while to protect the end of the word. This is called a hangover state. In such an SI device, transmission capacity and good sound quality are contradictory, so how to perform lockout control is a very important issue.

Conventional lockout control methods include a method of separating hungover channels and making them asymmetrical, a method of placing calls in a waiting state when calls are made all at once, that is, reducing lockout by inserting a buffer, and a voice puffing method. . The first method has the disadvantage that it cannot be controlled in conjunction with the audio detection level, and the latter two methods have the disadvantage that fluctuations occur in the audio.

The present invention provides a DSI lock-out control method that increases the voice detection level and shortens the hangover time in order to prevent sound quality deterioration due to lock-out, while controlling such that the sound quality deterioration due to the lock-out itself does not become large.

[Failure to solve the problem]

The present invention is characterized in that lockout control is performed by using a table that outputs the voice detection level and hangover time of the second voice detection section using the current lockout occurrence rate, voice detection level, and hangover time as parameters. This is achieved by providing a DSI lockout control scheme.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 shows D according to the DSI lockout control method of the present invention.
FIG. 2 is a block diagram of an SI device. Although not shown, the input signal 10 is a multiplexed digitized audio signal 9
M channel exists. The audio detector 1 is a detector that receives the audio signal and determines that there is a sound when the signal is above a certain level, and that there is no sound when the signal is below that level. For example, Figure 2 (al
As shown in the figure, in the case of an audio signal, there are two detection levels of the audio detector 1: a high level 100 and a low level 101. If the detection level is set to low level 101, the output of voice detector 1 will be low level 10 as shown in Figure 2 (bl).
Since the signal is input to the DST output channel from time ta corresponding to 1, in this case, the beginning of the voice is protected. Furthermore, if the detection level is set to a high level of 100, the output of the audio detector 1 will be input to the DST output channel from time t1 corresponding to the high level of 100 as shown in FIG. 2(C), so in this case , the beginning of the audio word will disappear. Also, regarding the ending of words, if you do not set hangover, the ending will be longer if you set it to a low level.

In FIG. 2(b), there is sound until time L3.

If the detection level is set to a high level, the ending of the word will be shortened, and the sound will disappear at time L2 in Figure 2 (C1).If the detection level is raised in this way, the sound period will be shortened, so the transmission capacity will be large. Therefore, the number of shut-out channels decreases.However, contradictory to this, the beginnings and endings of speech words are cut off, so subtle parts of the sound are lost and the sound quality deteriorates.Also, lowering the detection level improves the sound quality. On the other hand, as the voiced period becomes longer, the transmission capacity becomes smaller and the number of blocked channels increases.It is known that it is more unpleasant when the ending of a word is missed than the beginning of a word, so the detection level for the beginning of a word is higher than the detection level for the end of a word. The detection level is lowered so that the ending sound lags behind for a while.In other words, it creates a hangover state to protect the ending.The hangover time is shown in Figure 2 (in BL, time t3 is the starting point). In Fig. 2 (C), the starting point is large and the starting point is time t2, and in Fig. 2 (C) it is large, and in Fig. 2 (d+), the starting point is time L3, and it is small. In other words, when the current lockout rate is small, the hangover time is lengthened (to prevent deterioration of sound quality), and when the current lockout rate is large, the hangover time is shortened, the transmission capacity is increased, and the number of locked channels is increased. We are trying to reduce this.

In the block diagram of FIG.
counts the number t of channels that have become active from the output 11 of the audio detector 1 and is about to be transmitted to the output channel of the DSI device, and when t>N, the transmission allocation circuit 3 allocates N to control the allocation of N channels. It is output as a signal 31. At this time, by calculating t-N in the active channel counter 2, it is possible to know how many channels will be shut out, and therefore, the shut-out occurrence rate counter 4 is used to calculate the shut-out occurrence. Note that it is necessary to randomize the channels that are blocked out at this time so that they are not specific.

On the other hand, voice detection 2: (1) The current sound given as signal 50?IQ +l+ +7 bells is stored in the voice detection level register 5. The current hangover time is stored in a hangover time register 6, and this content is given as a signal 60 to the speech detector 1 to determine the length of the ending of the speech input 10. Become.

The present invention uses three parameters (A, B,
ROM for lockout control table with C) as address
(read-only memory) 7, and launch the next voice detection level 70 and next hangover time 71 to be output into the voice detection level register 5 and the hangover time register 6, respectively, by the enable signal 12 from the voice detection. It performs precise control to optimize the next voice detection level or hangover.

In this manner, the present invention performs lockout control using the three parameters of lockout rate, voice detection level, and hangover time, and thereby allows the manner in which the beginning and end of a word are truncated to be controlled separately. That is, the present invention uses the two parameters of voice detection level and hangover as table 7 determined by the current lockout occurrence rate, past voice detection level, and hangover time. By changing the contents of this table, the control section can obtain the best sound quality. Further, if detailed control is required, this can be done by increasing the capacity of the memory that stores the table.

〔Effect of the invention〕

According to the present invention, detailed control such as optimizing sound quality is possible by redifferentiating the table (increasing the capacity of the ROM).

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention. FIG. 2 is a waveform diagram showing the operational waveform of the output of the voice detector in the above embodiment, and FIG. 3 is a block diagram for generally explaining the DSI system. DESCRIPTION OF SYMBOLS 1...Speech detector, 2...Sound channel counter, 3...Transmission channel allocation circuit,
4... Lockout occurrence rate counter, 5
...Audio detection level register. 6... Hangover time register, 7... Lockout control table ROM. Patent Applicant: Fujitsu Limited, Friend, Patent Attorney
Hiroshi Matsuoka, TE 2nd figure ((j) tone color test (ondo detection and he"lshi; fm) (ha 2ri° over the limit; Jg) sound P product voice product 3rd figure

Claims (1)

[Claims] DSI lockout control, characterized in that lockout control is performed by using a table that outputs the voice detection level and hangover time of the next voice detection section using the current lockout occurrence rate, voice detection level, and hangover time as parameters. method.