Method of speech command recognition with dynamic assignment of probabilities according to the state of the controlled applications

A word recognition system designed to run as part of one program on a computer system which is capable of running a plurality of computer programs at one time, said word recognition system including:

• means for receiving user generated word signals representing words to be recognized;
• recognition means for performing pattern matching upon the word signals to select which one or more of a plurality of vocabulary words most probably correspond to each such word signal;
• other-program-monitoring means for repeatedly initiating transfer of command information from another program running on said computer including a set of commands which are currently active in said other program, which set varies over time;
• and
• probability-altering means responsive to said command information for increasing the probability that a word corresponding to one of the varying set of commands currently active at each of successive periods of time during the running of said other program will be selected by said recognition means as most probably corresponding to a received word signal during each such respective period of time.

A speech recognition system designed to run as part of one program on a computer system which is capable of running a plurality of computer programs at one time, said word recognition system including:

• means for receiving user generated acoustic word signals representing the spoken sound of words to be recognized;
• speech recognition means for performing pattern matching upon the acoustic word signals to select which one or more of a plurality of vocabulary words most probably correspond to each such acoustic word signal;
• other-program-monitoring means for repeatedly obtaining command information from another program running on said computer, which command information indicates which varying set of commands are currently active in said other program and what input has to be entered to cause said other program to execute each of said currently active commands;
• probability-altering means for repeatedly responding to said command information by increasing the probability that words corresponding to said currently active commands at each of successive time periods during the running of said other program will be selected by said recognition means as most probably corresponding to a received acoustic word signal during each such respective time period;
• and
• means for responding to the selection by said recognition means of one or more words corresponding to a given one of said commands by causing the command input associated with that given command to be supplied to said other program, so as to cause that other program to execute the given command.

A computerized method of performing speech recognition system in one program on a computer system which is capable of running a plurality of computer programs at one time, said method including the steps of:

• receiving user generated acoustic word signals representing the spoken sound of words to be recognized;
• performing pattern matching upon the acoustic word signals to select which one or more of a plurality of vocabulary words most probably correspond to each such acoustic word signal;
• repeatedly obtaining command information from another program running on said computer, which command information indicates a varying set of commands which are currently active in said other program;
• and
• repeatedly responding to said command information by increasing the probability that words corresponding to the varying set of currently active commands at each of successive time periods will be selected by pattern matching as most probably corresponding to a received acoustic word signal during each such respective time period.

A computerized method of performing speech recognition system in one program on a computer system which is capable of running a plurality of computer programs at one time, said method including the steps of:

• receiving user generated acoustic word signals representing the spoken sound of words to be recognized;
• performing pattern matching upon the acoustic word signals to select which one or more of a plurality of vocabulary words most probably correspond to each such acoustic word signal;
• repeatedly obtaining command information from another program running on said computer, which command information indicates which varying set of commands are currently active in said other program and what input has to be entered to cause said other program to execute each of said currently active commands;
• repeatedly responding to said command information by increasing the probability that words corresponding to said currently active commands at each of successive time periods in the running of said other program will be selected by pattern matching as most probably corresponding to a received acoustic word signal during each such respective time period;
• and
• responding to the selection by said recognition means of one or more words corresponding to a given one of said commands by causing the command input associated with that given command to be supplied to said other program, so as to cause that other program to execute the given command.

A computerized method of performing speech recognition on a computer system which has an operating system capable of running a plurality of computer programs at one time, said method including the steps of:

• receiving user generated acoustic word signals representing the spoken sound of words to be recognized;
• performing pattern matching upon the acoustic word signals to select which one or more of a plurality of vocabulary words most probably correspond to each such acoustic word signal;
• repeatedly using functions of said operating system to obtain time varying command information which identifies a set of one or more individual commands which are currently active in a given program at a given point in time and what input has to be entered to cause said given program to execute each of said individual currently active commands;
• repeatedly responding to said command information by increasing the probability that words corresponding to said currently active commands at each of successive time periods in the running of said program will be selected by pattern matching as most probably corresponding to a received acoustic word signal during each such respective time period;
• and
• responding to the selection by said recognition means of one or more words corresponding to a given one of said commands by causing the command input associated with that given command to be supplied to said given program through said operating system, so as to cause the given program to execute the given command.