JP2004007277A - Communication terminal equipment, sound recognition system and information access system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To use sound as an Internet input without increasing a cost of a terminal by transmitting the coding sound at a data transmission mode, recognizing the sound by using a transmitted sound parameter and inputting data by using the Internet in a sound recognition server. <P>SOLUTION: This equipment is provided with a sound input means inputting the sound and converting the inputted sound into a sound signal, a sound parameter extraction means extracting the prescribed sound parameter to the sound input means from the inputted sound, a data input means inputting data, a data transmission means transmitting data by using a procedure of transmission/reception on communication using a network for data communication, a sound transmission means transmitting the sound by using the procedure of transmission/reception on communication using the network for sound communication and a selection means selecting the data transmission means or the sound transmission means when the sound parameter is transmitted. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a communication terminal device that converts voice into character string data, a voice recognition system, and an information access system.
[0002]
[Prior art]
In recent years, Internet access using a mobile phone as a terminal has been widely used. The use of the Internet by a mobile phone involves inputting a keyword by making full use of a numeric keypad, and has a problem of poor usability.
[0003]
In order to solve the above-mentioned problem, it is conceivable to recognize speech by a terminal and use the speech as an input for using the Internet. However, it is not practical for the following reasons. First, with current speech recognition technology, free word recognition of unspecified speakers is not yet of sufficient quality. Further, mounting it on a mobile phone terminal requires a large increase in terminal cost.
[0004]
On the other hand, some systems for using voice as Internet input means have been proposed. A typical example is IBM's WebSphere Voice Server. This involves inputting voice using a telephone, converting the voice into a digital signal through A / D conversion, and transferring the digital signal to a voice server using an IP (Internet Protocol) network using VoIP (Voice over IP) technology. By performing speech recognition on a speech server and converting the text information into text information, the speech data is used as input data to the web server, and the text information output from the web server is converted into speech data using speech synthesis technology on the speech server. Is converted to a voice signal by VoIP technology via an IP network, and is output as voice using a telephone. The target phrase for speech recognition is not a free word but a limited one prepared in advance as a Web input, so that an unspecified speaker can be recognized.
[0005]
[Problems to be solved by the invention]
However, the WebSphere Voice Server system uses voice as an input for using the Internet, but makes it difficult to use a mobile phone as a terminal for the following reasons.
[0006]
First, there is the problem of voice quality. Normal dial calls and PHS (Personal Handy-phone System) are relatively high bit rate coding schemes such as 64 kbit / s PCM (Pulse Code Modulation) and 32 kbit / s ADPCM (Adaptive Differential PCM). Since digital encoding is used, the voice is less likely to deteriorate and high-precision voice recognition is possible. However, in the case of a PDC (Personal Digital Cellular) system or a CDMA system mobile phone which is currently most widely used, a 5.6 kbit / s CELP (Code Excited Linear Prediction) such as a VSELP (Vector Sum Excited Linear Prediction) system. Since encoding is performed using an encoding method, speech deterioration is more severe than PCM, ADPCM, or the like, and it is difficult to recognize speech with sufficient quality.
[0007]
In addition, a mobile phone uses a voice communication network in the case of a voice call (a procedure for transmission and reception relating to communication using the voice communication network is referred to as a “voice communication mode”), and in the case of an Internet connection. Performs communication using a packet network for data communication (a procedure for transmission and reception relating to communication using the network for data communication is referred to as “data communication mode”). In the case of data communication, since errors are not allowed, procedures for error correction and packet loss are clearly specified.However, since voice communication emphasizes real-time system, bits for error correction are added. Only transmission is performed, and measures against transmission path loss and the like are weaker than in data communication. This also makes speech recognition using mobile phone speech difficult.
[0008]
However, in the case of Internet input, unlike the case of normal voice communication, the demand for the real-time system is not so high. Further, the CELP-based speech coding scheme used in general cellular phones and cellular phones (excluding PHS) extracts parameters that can be used in speech recognition, such as linear prediction coefficients and pitch periods. Encoding voice using this encoding method and transmitting data only involves using the function originally installed in the portable terminal, so that there is almost no increase in cost.
[0009]
In view of the above circumstances, the present invention transmits coded voice in a data transmission mode, and a voice recognition server performs voice recognition using the transmitted voice parameters and uses the voice as input for Internet use, thereby increasing the cost of the terminal. It is intended to use voice as Internet input with little accompanying.
[0010]
[Means for Solving the Problems]
In order to achieve the above object, the invention according to claim 1 includes a voice input unit for inputting voice and converting the input voice into a voice signal, and extracting a predetermined voice parameter from the input voice to the voice input unit. Voice parameter extracting means, data input means for inputting data, data transmitting means for transmitting data using a transmission / reception procedure relating to communication using a data communication network, and communication using a voice communication network Voice transmission means for performing voice transmission using a transmission / reception procedure relating to the transmission of the voice parameter, and when the voice parameter is transmitted, a selection means for selecting either the data transmission means or the voice transmission means is provided. And
[0011]
According to a second aspect of the present invention, there is provided a receiving unit that receives a voice parameter, a voice recognition unit that performs a voice recognition process using the received parameter and outputs character string data, and a character string data output that transmits the character string data. Means.
[0012]
A third aspect of the present invention is characterized in that there is provided a means for converting an input voice into character string data using the first and second aspects of the invention.
[0013]
According to a fourth aspect of the present invention, character string data for input is obtained from a voice input using the voice recognition system according to the third aspect.
[0014]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
[0015]
FIG. 1 is a diagram showing a basic configuration of the information access system of the present invention. This configuration includes a first terminal device 100, a second terminal device 200, a Web server 300, a voice transmission network 301, a data transfer network 302, and the Internet 303.
[0016]
The first terminal device 100 includes a voice input unit 110, a voice parameter extraction unit 120, a data transmission unit 130, a voice transmission unit 140, a selection unit 150, and a data input unit 160.
[0017]
Further, the second terminal device 200 includes a receiving unit 210, a voice recognition unit 220, and a character string data output unit 230. In the following, the first terminal device is described as an example of a mobile communication terminal, particularly a mobile phone, but is not limited thereto.
[0018]
The second terminal device includes a receiving unit 210, a voice recognition unit 220, and a character string data output unit 230. The speech recognition means 220 uses a speech recognition algorithm using formant parameters such as LPC (Linear Predictive Coding) cepstrum known to experts.
[0019]
FIG. 2 is a diagram showing a basic operation of the first terminal device of the communication terminal device of the present invention. First, when making a normal voice call using the first terminal 100, a voice input unit 110, a voice parameter extraction unit 120, and a voice transmission unit 140 are used. The selection means 150 is set to use the voice transmission means 140.
[0020]
First, a voice is input to the voice input unit 110 and converted into a voice signal. Examples of the voice input unit 110 include a microphone device and a microphone device with a low-pass filter. If the audio input means 110 has an AD conversion function, the audio signal is a digital signal. If the audio input means 110 has no AD conversion function and the audio parameter extraction means 120 has an AD conversion function, the audio signal is It becomes an analog signal, but either may be used.
[0021]
The voice signal is input to a voice parameter extraction device 120, and voice parameters are extracted. The voice parameter is an encoding parameter for the mobile phone. The voice parameter may include an uncorrectable bit for error correction.
[0022]
The voice parameter is transmitted to the communication path by the voice transmitting unit 140. Here, a communication procedure for voice transmission (voice transmission mode) is used and transmitted to the other party 304 via the network 301 for voice transmission.
[0023]
FIG. 3 is a diagram showing an operation when Internet access is performed using the first terminal 100. When accessing the Internet using the first terminal 100, the data input unit 160 and the data transmission unit 130 are used. First, the user inputs data using the data input unit 160. As the data input means, in the case of a mobile phone terminal, a ten key for normal dialing can be used.
[0024]
Further, the input data is transmitted to the communication path by the data transmitting means 130. Here, a data transmission communication procedure (data transmission mode) is used, and the data is transmitted to a communication partner via the data transmission network 302 and the Internet 303.
[0025]
FIG. 4 is a diagram showing a configuration for Web access on the Internet by voice input. First, the operation of the first terminal device 100 will be described. In this case, a voice input unit 110, a voice parameter extraction unit 120, and a data transmission unit 130 are used. As in the case of making a voice call, a voice is input to the voice input means 110 and converted into a voice signal. Further, the audio signal is input to the audio parameter extracting means, and the audio parameter is output. In this case, an encoding method other than the encoding method used for the voice call may be used. However, considering that the cost of the terminal is greatly increased, the same encoding method as that for the voice call may be used. preferable. In speech recognition, since a formant parameter such as a LPC (Linear Predictive Coding) cepstrum coefficient is used, an encoding method for encoding a formant parameter such as an LPC coefficient, for example, a CELP-based encoding method may be used. Here, it is assumed that a CELP coding scheme is used.
[0026]
The extracted voice parameters are transmitted to the communication path by the data transmission means 130. Here, a data transmission communication procedure (data transmission mode) is used, and the data is transmitted to a communication destination via a data transmission network. For this reason, error-free transmission becomes possible.
[0027]
Next, the operation of the second terminal device 200 will be described. First, the voice parameter transmitted via the communication path is received by the receiving unit 210. The received voice parameters are input to the voice recognition means 220 and are converted into character string data. As the voice recognition means, since the voice parameters are extracted by using the CELP coding method, a part in which formant parameters such as LPC coefficients are coded is extracted therefrom and used for voice recognition.
[0028]
The character string data is transmitted as an input signal for Web access to the Web server 300 using a predetermined protocol, and the Web server performs processing based on the server settings.
[0029]
【The invention's effect】
As apparent from the above description, according to the present invention, input means for inputting voice and converting it into a voice signal, voice parameter extracting means for extracting a predetermined voice parameter from the voice input to the input means, data transmission Data transmission means for performing transmission using a mode, voice transmission means for performing transmission using a voice communication mode, further comprising: when transmitting the voice parameter, either the data transmission means or the voice transmission means Since the voice parameter is transmitted using the data transmission means, it is possible to transmit and use the voice parameter without error.
[0030]
Further, according to the present invention, receiving means for receiving the voice parameter, voice recognition means for performing a voice recognition process using the received parameter and outputting character string data, and a character for transmitting the character string data Since the apparatus includes the column data output means, by performing speech recognition using the error-free speech parameters, it is possible to obtain a high-quality speech recognition result using speech input from the communication terminal.
[0031]
Further, according to the present invention, the input speech is converted into character string data. Therefore, speech recognition is performed using the speech parameter without error, so that a speech recognition result of high quality can be obtained using speech input from a communication terminal. Can be obtained.
[0032]
Further, according to the present invention, since the input character string data is obtained from the voice input, information access using high-quality voice recognition becomes possible.
[Brief description of the drawings]
FIG. 1 is a diagram showing a basic configuration of an information access system of the present invention.
FIG. 2 is a diagram showing a basic operation of a first terminal device 100 of the communication terminal device of the present invention.
FIG. 3 is a diagram illustrating an operation in a case where an Internet access is performed using a first terminal 100;
FIG. 4 is a diagram showing a configuration for Web access on the Internet by voice input.
[Explanation of symbols]
100 first communication terminal device 110 voice input means 120 voice parameter extraction means 130 data transmission means 140 voice transmission means 150 selection means 160 data input means 200 second communication terminal device 210 reception means 220 voice recognition means 230 character string data output Means 300 Web server

Claims (4)

Voice input means for inputting voice and converting the input voice to a voice signal;
Voice parameter extraction means for extracting predetermined voice parameters from the input voice to the voice input means,
Data input means for inputting data;
A data transmission unit that performs data transmission using a transmission / reception procedure regarding communication using a data communication network,
Voice transmission means for performing voice transmission using a transmission and reception procedure related to communication using a voice communication network,
When transmitting the voice parameters,
A communication terminal device comprising a selection unit for selecting either the data transmission unit or the voice transmission unit. Receiving means for receiving voice parameters;
Voice recognition means for performing voice recognition processing using the received parameters and outputting character string data,
Character string data output means for transmitting the character string data. A speech recognition system comprising means for converting input speech into character string data using the communication terminal device according to claim 1 and the communication terminal device according to claim 2. 4. An information access system, wherein input character string data is obtained from voice input using the voice recognition system according to claim 3.

Images