Cancellation of loudspeaker words in speech recognition

A voice recognition system for use with a communication system having an incoming line and an outgoing line, said incoming line carrying an incoming signal from a first end to a second end operably attached to an audio output responsive to said incoming signal and said outgoing line carrying an outgoing signal from a second end to a first end, said outgoing line second end being attached to a microphone adjacent to said audio output and said outgoing line first end being attached to a variable system responsive to a selected command word in said outgoing signal, said voice recognition system comprising:

• a first speech recognition unit for detecting a selected outgoing word corresponding to said selected command word in said outgoing signal;
• a second speech recognition unit delayed relative to said first speech recognition unit for searching for an incoming word in said incoming signal corresponding to said selected outgoing word detected by said first speech recognition unit during said delay;
• and
• a signal generator operably connected to said second speech recognition unit, said signal generator generating a command corresponding to said selected command word when said incoming word corresponding to said selected outgoing word is not detected by said second speech recognition unit.

A method for speech recognition in a communication environment where both directions of said communication system are active at the same time, said method comprising:

• applying speech recognition to an incoming signal to detect an incoming word;
• applying speech recognition to an outgoing signal to detect an outgoing word;
• outputting a command signal associated with a detected selected outgoing word which does not match said incoming word;
• and
• wherein said applying speech recognition to said incoming and outgoing signals results in the detection of a null word in the incoming signal and an outgoing word in the outgoing signal resulting in the output of said outgoing word.

A method for speech recognition in a communication environment where both directions of said communication system are active at the same time, said communication system having a near end and a far end, an incoming line carrying an incoming signal from the far end to the near end of said communication system, an outgoing line carrying an outgoing signal from the near end to the far end of said communication system, a speaker operably attached to the far end of said incoming line for audio output of said incoming signal, a microphone operably attached to the near end of said outgoing line for picking up sound for inclusion in said outgoing signal, said method comprising:

• applying speech recognition to said incoming signal in said incoming line to detect an incoming word;
• applying speech recognition to said outgoing signal in said outgoing line to detect an outgoing word;
• comparing said outgoing word to said incoming word;
• and
• outputting a valid word when the outgoing word does not match the incoming word.