Device control device, speech recognition device, agent device, on-vehicle device control device, navigation device, audio device, device control method, speech recognition method, agent processing method, on-vehicle device control method, navigation method, and audio device control method, and program

A device control method, comprising:

• performing, in a computer:
• a speech recognition step of acquiring speech data representing a speech and specifying words candidates included in the speech by performing speech recognition on the speech data and calculating a likelihood of each of the specified words candidates;
• a specifying step of specifying words included in the speech based on likelihoods specified in the speech recognition step and specifying a content of the speech uttered by an utterer based on the specified words;
• and
• a process execution step of specifying content of a subsequent control to be performed on an external device to be a control target based on a currently executed control, a predetermined subsequent control, a weighting factor associated with the currently executed control, and the specified content of the uttered speech, and performing the subsequent control, wherein the process execution step obtains the weighting factor by calculating a product of transition constants defined on routes from the currently executed control to the subsequent control associated with the currently executed control, and, among the subsequent controls associated with the currently executed control, identifies a control in which a product is a largest product of the weighting factor and the calculated likelihood.

A non-transitory recording medium storing a program which allows a computer to function as:

• speech recognition means which acquires speech data representing a speech and specifies words candidates included in the speech by performing speech recognition on the speech data and calculates a likelihood of each of the specified words candidates;
• specifying means which specifies words included in the speech based on likelihoods calculated by the speech recognition means and specifies a content of the speech uttered by an utterer based on the specified words
• a database which stores preceding controls, subsequent controls, and weighting factors, each of which is associated with one another therein;
• and
• process execution means which specifies content of a subsequent control to be performed on an external device to be a control target based on a currently executed control, a weighting factor stored in association with the currently executed control, and the content of the uttered speech specified by the specifying means, and performs the subsequent control, wherein the process execution means obtains the weighting factor by calculating a product of transition constants defined on routes from the currently executed control to the subsequent control associated with the currently executed control, writes the obtained weighting factor into the database, and, among the subsequent controls stored in the database associated with the currently executed control, identifies a control in which a product is a largest product of the weighting factor and the calculated likelihood.