Voice recognizing apparatus and voice recognizing method

A voice recognizing apparatus, comprising

• circuitry configured to:
• receive an electrical signal as a voice signal and an electrical signal as a noise signal, the electrical signal as the voice signal being converted from voice of a talker, the electrical signal as the noise signal being converted from noise in usage environment;
• calculate, as a voice level, a level of the received voice signal of the talker;
• calculate, as a noise level, a level of the received noise signal;
• perform conversion of a waveform of the electric signal as the voice signal into a character string;
• calculate reliability of the conversion when the conversion of the waveform into the character string is performed;
• and
• calculate a necessary voice level on a basis of the voice level, the noise level, and the reliability, and
• a necessary reliability database configured to hold necessary reliability that is required for each operation,
• wherein the circuitry is configured to:
• estimate an operation of the talker from the character string after the conversion;
• and
• execute the operation on a basis of the estimated operation, wherein
• the circuitry is configured to compare the reliability with the corresponding necessary reliability stored in the necessary reliability database, and not to execute the estimated operation in a case where the reliability does not exceed the necessary reliability.

A voice recognizing method, comprising:

• receiving voice of a talker and noise in usage environment;
• performing conversion of the received voice of the talker into an electric signal as a voice signal and conversion of the received noise into an electrical signal as a noise signal;
• calculating, as a voice level, a level of the voice signal of the talker;
• calculating, as a noise level, a level of the noise signal;
• performing conversion of a waveform of the electric signal as the voice signal into a character string;
• calculating reliability of the conversion of the waveform into the character string when the conversion of the waveform into the character string is performed, wherein
• a basic reliability of a specific talker in a situation where the noise level is low is calculated;
• and
• calculating a necessary voice level on a basis of the voice level, the noise level, and the reliability, wherein
• the necessary voice level is corrected on a basis of the basic reliability.