JPS61126855A - Voice dial controlling system - Google Patents

Abstract

PURPOSE:To reduce the malrecognition of the voice dial information caused by the transmission characteristic deviation by correcting the voice dial information transferred from the calling telephone system which is performed based on the characteristic deviation of the transmission line between the calling telephone set and the voice recognizing means from that of the standard transmission line. CONSTITUTION:Prior to receiving the voice dial information, a voice recognizing circuit 6 transmits a reference voice information (c) from the calling telephone system 2. Based on the recognition result of the voice information (c) and the reference message standard pattern 641, the circuit 6 corrects the recognition result by its characteristic correcting part 621, and later, starts the receiving of the voice dial information. Thus, the voice dial information e1-en transferred from the telephone 2 having different transmission characteristics are corrected to an approximate standard level, and hence the malrecognition is reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a voice dialing control system in an automatic exchange capable of receiving voice dialing information.

In recent years, with the development of voice recognition technology, attempts have been made to use the voice of the caller in the dial information sent by the calling telephone to select the called telephone in an automatic exchange.

FIG. 3 is a diagram showing an example of an automatic exchange to which the present invention is applied. In FIG. 3, when a caller makes a call from a telephone 2 housed in an automatic exchange 1 to a telephone 3, a central controller 4 connects the calling telephone 2 to a voice recognition circuit 6 via a network 5. Connect to. In this state, when the calling party speaks the telephone number assigned to the called telephone 3, the voice recognition circuit 6 recognizes the voice dialing information transmitted from the telephone 2 via the network 5, and the recognition result, i.e. The telephone number of the called telephone 3 is transmitted to the central control unit 4 . The central control device 4 calls the called telephone 3 based on a known method, and when it answers, the calling telephone 2 and the network 5 are connected.
Connect via.

If the voice recognition circuit 6 erroneously recognizes the voice dial information, the called telephone 3 may be disconnected or disconnected. Therefore, in order for the automatic exchange 1 to provide a stable exchange connection, the voice dial It is desired to improve the information recognition rate.

[Conventional technology]

FIG. 4 is a diagram showing an example of a conventional voice recognition circuit in this type of automatic exchange.

In FIG. 4, voice dial information transmitted from a calling telephone (not shown) is converted into a digital format by an analog-to-digital converter 61 and transmitted to an analysis feature extractor 62. The analysis feature extraction unit 62 divides the transmitted voice dial information into a predetermined number (for example, 16) of frequency bands, detects the reception level in each frequency band, and transmits the detection result to the verification unit 63. The standard pattern generation unit 64 stores in advance signal levels (hereinafter referred to as standard patterns) in each frequency band regarding a plurality of standard voice dial information transmitted via a transmission path having standard transmission characteristics. has been done.

The collation unit 63 compares the received level in each frequency band transmitted from the analysis feature extraction unit 62 with the standard pattern generation unit 6.
4, and the voice dial information corresponding to the most similar standard pattern is output as a recognition result and transmitted to a central control device (not shown). Note that the control section 65 controls the voice dial information recognition operation in the analog-to-digital conversion section 61 to the standard pattern generation section 64.

Note that the characteristics of the transmission path between the telephone 2 housed in the automatic exchange 1 and the voice recognition circuit 6 are diversified. For example, the fifth
As shown in the figure, the standard loss characteristics assumed when setting the standard pattern 1. If the loss characteristic Lx of the transmission line corresponding to the calling telephone 2 is different from s, the amplitude characteristic Ax detected by the analysis feature extracting section 62 is different from the amplitude characteristic As of the standard pattern held by the standard pattern generating section 64. As shown in FIG. 6, a deviation occurs, causing erroneous recognition.

[Problem that the invention seeks to solve]

As is clear from the above explanation (in conventional voice dialing control systems, a standard pattern set for a transmission line with standard characteristics was adopted as a target for matching. Therefore, various transmission characteristics were There is a risk that the recognition result of voice dialing information transmitted from a telephone equipped with a telephone may be erroneously recognized due to deviations in transmission characteristics.

In order to eliminate such problems, it may be considered to store a plurality of recognition results of voice dialing information sent from telephones 2 having various transmission characteristics in the standard pattern generating section 64; It is also necessary to go to the location where each telephone in the exchange network is installed and speak.
It required a lot of effort.

[Means for solving problems]

The problem is that in an automatic exchange equipped with voice recognition means for identifying voice dialing information sent from a telephone, the voice recognition means recognizes a predetermined reference voice transmitted from the telephone before receiving the voice dialing information. The problem is solved by the present invention, which is characterized in that information is received, the obtained recognition result is compared with a pre-held reference recognition result, and the voice dialing information transmitted from the telephone is corrected based on the comparison result.

[Effect]

That is, according to the present invention, before transmitting the voice dialing information, the known reference voice information transmitted from the calling telephone sets the transmission path between the calling telephone and the voice recognition means to be different from the standard transmission line. Since the deviation in the characteristics between the transmission characteristics is ascertained and the voice dialing information transmitted from the calling telephone is corrected based on the deviation, misrecognition of the voice dialing information due to the deviation in the transmission characteristics is reduced.

Note that the reference voice information may be sent by the caller speaking the reference message or automatically sent by the telephone.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram illustrating a speech recognition circuit according to an embodiment of the present invention, and FIG. 2 is a diagram illustrating a connection process according to an embodiment of the present invention. Note that the same reference numerals indicate the same objects throughout the figures. The target automatic switchboard is shown in Figure 3.

In FIG. 1, the analysis feature extraction section 62' includes a characteristic correction section 621 as means for correcting the voice dial information transmitted from the calling telephone, and a standard pattern generation section 64''.
is a reference speech information standard pattern 64 obtained when predetermined reference speech information is recognized via standard transmission characteristics.
1 is the standard recognition result, and the control section 65'' is further provided with a characteristic correction section 621 and a correction control section 651 that controls the correction operation based on the reference message standard pattern 641, respectively.

In FIGS. 1 to 3, when a caller sends a calling signal a to the automatic exchange 1 in order to make a call from a telephone 2 housed in the automatic exchange 1 to a telephone 3, the central controller 4 The calling telephone 2 is connected to the voice recognition circuit 6 via the
and sends the caller a standard message (e.g. "
A standard message request signal b (for example, "Before speaking the called party's telephone number, the standard message "A, I, T, O") is requested to be uttered.・Engineering・
"Please speak to me") is sent back to telephone 2. When the caller who received the standard message request signal utters the standard message "A, I, Tsu, Engineering, O", the telephone 2
transmits the reference speech information C corresponding to the reference message to the speech recognition circuit 6. In the speech recognition circuit 6, the analog-to-digital conversion section 61 converts the received reference speech information C into a digital format and transmits it to the analysis feature extraction section 62', and the analysis feature extraction section 62' converts the received reference speech information C into digital format. The received levels in a predetermined number (for example, 16) of frequency bands are detected and transmitted to the matching section 63'. Collation section 6
3° is the received level in each frequency band transmitted from the analysis feature extractor 62", and the standard pattern generator 64°
The deviation value is compared with the reference message standard pattern 641 extracted from the standard message pattern 641, and the deviation value is transmitted to the control unit 65° as a correction value. In the control unit 651, the correction control unit 651
Based on the correction value transmitted from the analysis feature extraction unit 62
After correcting the attenuation characteristics of each frequency band by the characteristic correction unit 621 in A signal d is returned. When the caller who has received the dialable signal d speaks the telephone number assigned to the called telephone 3, the telephone 2 performs voice recognition on the voice dialing information e1 to en corresponding to the utterance. The voice recognition circuit 6 recognizes the voice dial information e1 to en transmitted from the telephone 2 via the network 5 by the analysis feature extraction section 62' corrected by the characteristic correction section 621 and the collation section 63°. and transmits the recognition result, that is, the telephone number of the called telephone 3, to the central control device 4.The central control device 4 sends a ringing signal f to the called telephone 3 based on a known method.
When the called telephone 3 responds and sends back a response signal, the transmission of the ringing signal f and the ring tone g is stopped, and the calling telephone 2 and the called telephone A telephone 3 is connected via a network 5, and a call i is started.

As is clear from the above description (according to this embodiment, the voice recognition circuit 6 causes the calling telephone 2 to send out the reference voice information C before receiving the voice dial information, and the recognition result of the reference voice information C is After correction is performed by the characteristic correction unit 621 using the reference message standard pattern 641 and the standard message pattern 641, reception of voice dial information is started.Therefore, the voice dial information e transmitted from each telephone 2 having different transmission characteristics.
1 to en are corrected to the road sign rate level, reducing misrecognition.

Note that FIGS. 1 to 3 are only one embodiment of the present invention, and for example, the reference voice information C is not limited to that based on the reference message uttered by the caller, and the telephone 2 is equipped with a function to send known reference voice information C in advance, and when a call is made, it is connected to the voice recognition circuit 6, and after confirming that the reference message request signal C is returned, the reference voice information is sent to the function. Although it is also possible to automatically send out the information C, the effects of the present invention do not change even in such a case.

Furthermore, the configuration of the voice recognition circuit 6 is not limited to that shown in the drawings (although many other modifications may be considered, but the effects of the present invention remain the same in any case.

〔Effect of the invention〕

As described above, according to the present invention, in the automatic exchange, before transmitting the voice dial information, the transmission path between the calling telephone and the voice recognition means is controlled by the known reference voice information sent from the calling telephone. The deviation in characteristics from the standard transmission path is ascertained, and the voice dial information transmitted from the calling telephone is corrected based on the deviation, which reduces misrecognition of voice dial information due to deviations in transmission characteristics. .

[Brief explanation of drawings]

FIG. 1 is a diagram showing a speech recognition circuit according to an embodiment of the present invention, FIG. 2 is a diagram showing a connection process according to an embodiment of the present invention,
FIG. 3 is a diagram showing an example of an automatic exchange to which the present invention is applied, FIG. 4 is a diagram showing an example of a conventional voice recognition circuit, and FIG. 5 is a diagram showing an example of a conventional voice recognition circuit.
FIG. 6 is a diagram showing an example of deviation in loss characteristics of a transmission path, and FIG. 6 is a diagram showing an example of deviation in amplitude characteristics of audio information. In the figure, 1 is an automatic exchange, 2 and 3 are telephones, and 4
5 is a central control unit, 5 is a network, 6 is a voice recognition circuit, 61 is an analog-to-digital converter, 62 and 6 are
2" is an analysis feature extraction section, 63 and 63" are matching sections, 6
4 and 64' are standard pattern generators, 65 and 65
° is a control unit, 621 is a characteristic correction unit, 641 is a standard message standard pattern, 651 is a correction control unit, a is a calling signal, b is a standard message request signal, C is standard voice information, d
is a dialable signal, el to en are voice dialing information,
f is a ringing signal, g is a ringing tone, h is a response signal, i is a call,
shows. Arithmetic 1 Zosen 2 Sendai 4 Tadashi

Claims (3)

[Claims] (1) In an automatic exchange equipped with voice recognition means for identifying voice dialing information transmitted from a telephone set, prior to receiving the voice dialing information, the voice recognition means recognizes predetermined reference voice information transmitted from the telephone set. A voice dial control method characterized in that the received recognition result is compared with a pre-held reference recognition result, and voice dial information transmitted from the telephone is corrected based on the comparison result. (2) The voice dialing control system according to claim 1, wherein the reference voice information is sent from the telephone according to voice uttered by a caller based on a request from the voice recognition means. (3) The voice dialing control system according to claim 1, wherein the reference voice information is automatically sent out when the telephone is connected to the voice recognition means.