JPH03123156A - Voice dial equipment - Google Patents

Abstract

PURPOSE:To take the interchangeability of a voice standard pattern by correcting a frequency characteristic of plural input devices. CONSTITUTION:The equipment is provided with a feature extracting part 11 for extracting the feature quantity of a voice with respect to the name inputted by the voice, input devices of two systems of a handset 1 and a hand-free microphone 2, and a correction filter 3 for allowing a frequency characteristic of the handset 1 to be approximate to a frequency characteristic of the hand-free microphone 2. In such a state, an output of the hand-free microphone 2, and an output of the handset 1 passing through the correction filter 3 are connected selectively to an input of the feature extracting part 11. In such a way, frequency characteristics of plural input devices are corrected roughly equally, and also, the input device can be designated selectively, therefore, a drop of a recognition rate is not caused, and the interchangeability by the input device of the standard pattern is taken.

Description

[Detailed description of the invention] Kanchibu Station The present invention relates to a voice dialing device.

As is well known, prior art voice dialing devices do not require manual dialing when making a call, but automatically output a dialing signal based on voice input. Figure 3 shows the configuration of a typical voice dialing device. Give an example.

In FIG. 3, 10 is a transmitter, 11 is a feature extraction unit, and 1
2 is a standard pattern for synthesis, 13 is a speech synthesis section, 14 is a data control section, 15 is a keyboard, 16 is a display section, 17 is a standard pattern for recognition, 18 is a pattern matching section,
19 is a number storage unit, 20 is a transmission circuit, 21 is a receiver, 2
2 is a telephone communication circuit, 23 is a hook switch, 24 is a circuit control unit, and 25 is a central office line.The voice input from the transmitter 10 is extracted by a feature extraction unit 11, and is prepared in advance. is matched against a standard pattern. As a result of step 3, the telephone number corresponding to the standard pattern with the highest degree of similarity is output. Further, the speech synthesis unit 13 in the figure is used for the purpose of confirming the other party and outputting operation guidance.

The feature quantity is the relative value of the bandpass filter output sampled in the time series (time, spectrum, etc.).
pattern) or bandpass filter instead of LPG or L
Most of them use the PC cepstrum, and all of them focus on the frequency characteristics of the voice.

When using the above configuration and feature amounts, the frequency characteristics of the transmitter 10 must be the same when creating the standard pattern and when dialing by voice to an extent that does not cause misrecognition. There is no problem if a single handset is always used as a talker.

In addition, if there are two or more transmitters, such as a telephone equipped with a hands-free microphone, and the frequency characteristics are significantly different, it is necessary to limit voice dialing to one transmitter, which may result in erroneous recognition. Cause.

- Purpose The present invention was made in view of the above-mentioned circumstances, and
In particular, this was done for the purpose of correcting the frequency characteristics of a plurality of input devices to ensure compatibility of standard audio patterns.

In order to achieve the above-mentioned object, the present invention includes a feature extraction unit that extracts the feature quantity of the voice for a name input by voice, and a feature quantity extracted by the feature extraction unit. A standard pattern storage unit that stores the standard pattern; a number storage unit that stores the dial number corresponding to the name input by voice; and the feature amount extracted by the feature extraction unit during voice input and the standard stored in advance. a pattern matching unit that recognizes which standard pattern the input voice corresponds to by performing pattern matching with feature quantities in the pattern storage unit;
A voice dialing device comprising a dialing circuit that reads out a dialing number of a name corresponding to the recognition result by the pattern matching section from the number storage section and outputs a dialing signal. It has a plurality of correction filters that bring the frequency characteristics closer to each other, and the outputs of the plurality of correction filters are selectively connected to the input of the feature extraction section, or the first and second two systems. It has an input device and a correction filter that approximates the frequency characteristics of the second input device to the frequency characteristics of the first input device, and the output of the first input device and the second The present invention is characterized in that the output of the input device is selectively connected to the input of the feature extraction section. Hereinafter, the present invention will be explained based on examples.

FIG. 1 is a configuration diagram for explaining one embodiment of the present invention. In the figure, 1 is a handset, 2 is a hands-free microphone,
3 is a correction filter, 4 is a switching section, and other parts having the same functions as the conventional device shown in FIG. 3 are given the same reference numerals as in the case of FIG. 3. The embodiment shown in FIG. 1 shows a configuration example in which the input device has two systems, a handset and a hands-free microphone, and the frequency characteristic of the handset 15- is shown by the solid line A in FIG. 2(a). , the characteristics of the hands-free microphone 2 are shown by the dashed line B in the figure.

The output of the handset passes through a correction filter having the characteristics shown by curve C in FIG.
c) - Corrected so as to approximate the characteristics of the hands-free microphone indicated by the dashed line E.

The user uses a key to specify whether to use the handset or hands-free microphone for voice input. This designation result is realized by switching the SW of the switching section by the data control section. Since the characteristics of each microphone are corrected almost equally as described above, the standard pattern can be created using either microphone, and the pattern will be the same during dialing.

Further, in this embodiment, there are two systems of input devices, and one of the correction filters is omitted, but the present invention does not limit the number of input devices, the number of correction filters, the configuration, etc.

As is clear from the above explanation, according to the present invention, the frequency characteristics of a plurality of input devices can be corrected almost equally, and input devices can be selectively specified, without causing a decrease in recognition rate. , compatible with standard pattern input devices.

[Brief explanation of drawings]

FIG. 1 is a block diagram for explaining one embodiment of the present invention, FIG. 2 is a diagram for explaining the operation of the present invention, and FIG. 3 is an example of a conventional voice dialing device. FIG. 1...Handset, 2...Hands-free microphone,
3... Correction filter, 4... Switching section, 10... Transmitter, 11... Feature extraction section, 12... Standard pattern for synthesis, 13... Speech synthesis section, 14...・Data control section, 15... Keyboard, 16... Display section, 17'... Standard pattern for recognition, 18... Pattern matching section, 19... Number storage section, 20... Transmission circuit , 21... receiver, 22... communication circuit, 23...
Hook switch, 24...8- circuit control section, 25... station line.

Claims (1)

[Scope of Claims] 1. A feature extraction unit that extracts features of a name input by voice, and a standard pattern storage that stores the features extracted by the feature extraction unit as a standard pattern. a number storage section that stores a dial number corresponding to the name inputted by voice, and a feature amount extracted by the feature extraction section during voice input and a feature amount stored in the standard pattern storage section stored in advance. A pattern matching unit that recognizes which standard pattern the input voice corresponds to by performing pattern matching, and a dial number of the name corresponding to the recognition result by the pattern matching unit is read from the number storage unit and outputs a dial signal. A voice dialing device having a plurality of input devices and a plurality of correction filters that bring frequency characteristics specific to the plurality of input devices closer to each other, and selectively outputs the outputs of the plurality of correction filters from the above. A voice dialing device characterized in that it is connected to an input of a feature extraction section. 2. A feature extraction unit that extracts the feature amount of the voice input for the name input by voice, a standard pattern storage unit that stores the feature amount extracted by the feature extraction unit as a standard pattern, and a name input by voice. By performing pattern matching between the number storage section that stores the dial number corresponding to the name being entered, and the feature amount extracted by the feature extraction section during voice input and the feature amount stored in the standard pattern storage section that is stored in advance. Equipped with a pattern matching unit that recognizes which standard pattern the input voice corresponds to, and a transmission circuit that reads out the dial number of the name corresponding to the recognition result by the pattern matching unit from the number storage unit and outputs a dial signal. A voice dialing device comprising two systems of input devices, first and second, and a correction filter that approximates the frequency characteristics of the second input device to the frequency characteristics of the first input device,
the output of the input device and the second input device via the correction filter.
A voice dialing device characterized in that the output of the input device is selectively connected to the input of the feature extraction section.