JPS63121098A - Voice recognition equipment for telephone - 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 Field of the Invention The present invention relates to a telephone speech recognition device that recognizes speech input from a telephone terminal.

2. Description of the Related Art In recent years, voice recognition devices for recognizing telephone voices have gradually come into general use.

A conventional telephone voice recognition device will be described below with reference to the drawings. FIG. 2 shows a conventional telephone voice recognition device. In FIG. 2, 10 and 11 are various telephone terminals, each of which is connected to the selection means 3 in the telephone exchange 2. In FIG. Furthermore, the output of the selection means 3 is input to the voice matching section 4. On the other hand, the output of the standard pattern set 5 is also connected to the voice verification section 4.

The operation of the telephone voice recognition device configured as described above will be described below.

The user inputs voice from telephone terminal 1Q or 11,
These voices are input to the telephone exchange 2. A selection means 3 in the telephone exchange 2 determines from which telephone terminal the voice input is received, controls a switch, and outputs the voice to be recognized to the voice matching section 4. On the other hand, in the standard pattern set 5, a set of statistically optimal speech analysis results for each recognition target speech is stored, and these data are also input to the speech matching section 4. The voice verification section receives the above-mentioned two systems of input, performs a predetermined verification process, and outputs a recognition result. Many methods have been devised for this matching task, but most of them are based on the concept of spectral distance between the input speech and the standard pattern, and are based on the drift pattern that gives the closest spectral distance. The voice is used as the recognition result of the input voice.

Problems to be Solved by the Invention However, in the above configuration, when the telephone terminals used by the users have different acoustic characteristics and different spectra are given to the voices, the voice verification unit 4
In the comparison of spectral distances carried out in , there was a problem in that the spectral distances due to differences in telephone terminals made the spectral distance differences due to differences in voice itself unclear, leading to erroneous recognition.

- As an example, FIG. 3 shows the frequency characteristics of the transmitting system of three types of commercially available telephone terminals. FIG. 3(L) shows the characteristics of a transmission system using a carbon type microphone, and the wide peak characteristics at 2 to 3KH and the shoulder characteristics at 300H7 are noticeable. On the other hand, FIG. 3(b) shows the characteristics of a transmitting system using an electret condenser microphone, and the remarkable characteristics seen in FIG. 3(&) do not appear in this. Furthermore, FIG. 3(C) shows the characteristics of another transmission system using the same electret condenser microphone, and only a wide peak characteristic at 2 to 3 kHz appears. As described above, the frequency characteristics of the transmitting system of telephone terminals vary greatly depending on the model. Therefore, it is easy to imagine that the frequency customization of the audio input from these devices also differs greatly. Furthermore, it is easy to imagine that if voices input from different models of telephone terminals are compared using a single type of standard pattern, the differences in the spectral distances of the voices themselves will become unclear, leading to misrecognition. There is a problem with this.

In view of the above-mentioned problems, the present invention provides a telephone speech recognition device that can perform good speech recognition even when the acoustic characteristics of telephone terminals used by users are different from each other.

Means for Solving the Problems In order to achieve this objective, the telephone voice recognition device of the present invention identifies and selects a plurality of telephone terminals and which telephone terminal the input voice comes from. a selection means, a filter means for filtering the output of the selection means, a set of standard patterns, a plurality of ranges of filter coefficients to be input to the filter means, a plurality of telephone terminals and a plurality of sets of the filter coefficients; It consists of a correspondence table that records a predetermined many-to-one correspondence between the two, and a voice matching section.

Effect: With this configuration, the telephone terminal into which the speech to be recognized is input is determined, and one set of filter coefficients corresponding to the acoustic characteristics of this telephone terminal is determined. Here, each set of filter coefficients is given a spectrum in advance that corresponds to the acoustic characteristics of the telephone terminals associated with it in the correspondence table, thereby achieving good speech recognition that cancels out differences between telephone terminals. It can be done.

EXAMPLE Hereinafter, an example of the present invention will be described with reference to one drawing.

FIG. 1 shows a telephone voice recognition device according to an embodiment of the present invention. In FIG. 1, 10 and 11 are customer telephone terminals, and these are lines 60 and 6, respectively.
Telephone exchange after 1? It is connected to the selection means 3 in A2. Further, the output of the selection means 3 is applied to the filter means 7, and the output of the filter means 7 is inputted to the voice verification section 4. The selection information of the - way selection means 3 is input to a correspondence table 9, and the table index result output of the correspondence table 9 is input to a plurality of sets 8 of filter coefficients. Furthermore, one set of filter coefficients selected from the plurality of sets 8 of filter coefficients is input to the filter means 7. Further, the contents of the standard pattern set 5 are input to the voice matching section 4.

Regarding the telephone voice recognition device configured as above,
The operation will be explained below.

A user inputs voice from the telephone terminal 1o or 11, and these voices are input to the telephone exchange 2 via lines 60 and 61. The selection means 3 in the telephone exchange 2 determines from which line the voice input is received and outputs the voice to be recognized to the filter means 7 by controlling a switch. For convenience of explanation, the telephone terminal used by the user for voice input will be referred to as an active terminal, and the line to which it is connected will be referred to as an active line. The selection means 3 outputs the line number of the current anonymous line to the correspondence table 9. The correspondence table uses this active line number to search a many-to-one table, determines the type of active terminal, and further cancels out the acoustic characteristics of the active terminal imparted to the audio signal at the input stage of the filter means 7. The set of filter coefficients most suitable for The selected set of filter coefficients is input to filter means 7, and filter processing therein yields an audio signal in which the acoustic characteristics of the active terminal are canceled. The voice verification section 4 receives this voice signal and the output from the standard pattern set 5, and performs voice verification.

However, in the present invention, a set of filter coefficients that cancel out the acoustic characteristics of each telephone terminal must be created in advance. - For example, there is a method of measuring the transmitting impulse response of each type of telephone terminal in advance and configuring a digital filter that approximates the inverse characteristics of the impulse response. With this method, under the assumption of a linear system, it is possible to completely correct the transmission system of the telephone terminal within the range of the above approximation. This method is suitable when an electret condenser microphone is used in the transmission system. On the other hand, when a carbon type microphone is used, due to its nonlinearity, the above-mentioned correction based on the inverse characteristic cannot be strictly achieved. However, as a rough approximation, it is considered that the desired objective can be achieved by measuring the typical frequency characteristics of the target carbon microphone and applying similar processing to this.

As described above, according to the embodiment, a signal with the characteristics of the telephone terminal canceled out is always obtained as the output of the filter means 7, regardless of the telephone terminal used. Voice verification can be performed. Furthermore, even if the lines 6o or 61 in FIG. 1 have specific different acoustic characteristics, this shadow 9 can be canceled out or significantly reduced by the same method.

Effects of the Invention As described above, the present invention provides a plurality of telephone terminals, a selection means for determining and selecting which telephone terminal the input voice comes from, and a filter means for filtering the output of the selection means. and a predetermined many-to-one correspondence relationship between a set of standard patterns, a plurality of sets of filter coefficients to be input to the filter means, and the plurality of telephone terminals and the plurality of sets of filter coefficients. By providing a correspondence table for storing the information and a voice matching section, it is possible to perform good voice recognition that cancels out differences in the acoustic characteristics of the telephone terminal models to which the voice is actually input, and furthermore, the selection means Since the terminal and line are identified and selected as a pair, we have developed an excellent telephone voice recognition device that can perform good voice recognition using the same method even when the line has specific different sound characteristics. This is something that can be achieved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of a telephone voice recognition device according to an embodiment of the present invention, FIG. 2 is a block diagram of a conventional telephone voice recognition device, and FIG. 3 (λ), (b), and (C) are It is a frequency characteristic diagram of the transmitting system of three commercially available telephone terminals. IQ, 11...Telephone terminal, 2...Telephone exchange, 3...Selecting means, 4...Voice matching section, 6...Q ? ! The ancestor of the g pattern, 60.6
1... Line, 7... Filter means, 8
...Multiple sets of filter coefficients, 9...Correspondence table. Name of agent: Patent attorney Toshio Nakao and one other person
- Figure 2 Figure 3 (dB+ Utaha Chichi ()lzJ (dB+ □Thinking thing Hυ -m friend? ~ (92)

Claims (1)

[Claims] a plurality of telephone terminals, a selection means for determining and selecting from which telephone terminal the input voice comes from, a filter means for filtering the output of the selection means, a set of standard patterns, and the filter means. a correspondence table that stores a predetermined many-to-one correspondence relationship between a plurality of sets of filter coefficients to be input into a plurality of telephone terminals and a plurality of sets of filter coefficients; The selection means determines and selects from which telephone terminal the voice to be recognized that is input from any one of the plurality of telephone terminals is input, and the determination and selection result A corresponding set of filter coefficients is selected from the plurality of sets of filter coefficients based on the correspondence table, and the audio input to the filter means is processed using the selected set of filter coefficients. A voice recognition device for a telephone, characterized in that filter processing is performed and the output is compared with the set of standard patterns by the voice matching unit.