Distributed environmental microphones to minimize noise during speech recognition

A method to synchronize one or more mutually non-synchronous audio signals, the method comprising:

• at a user microphone of an electronic system, said user microphone configured to be worn by a user or collocated with a speech organ of the user, receiving a first audio signal representative of a first sound in the environment occupied by the user;
• sending from the user microphone to a coupled hardware processor of the electronic system a first audio signal representative of the first sound;
• at a plurality of environmental microphones of the electronic system configured to be positioned remotely from the user and distributed at different locations in an environment occupied by the user, detecting a second sound in the environment occupied by the user, wherein at least one environmental microphone of the plurality of environmental microphones is configured to be positioned at a location such that the detected second sound excludes the first sound;
• and
• via a transmitter of the electronic system coupled with each environmental microphone of the plurality of environmental microphones, transmitting to the hardware processor a second audio signal representative of the detected second sound, wherein:
• the first audio signal and the transmitted second audio signal are representative of a common, distinctive environmental sound event emanating from a single environmental source, but are mutually non-synchronous due to the spatial separation of the user microphone and each environmental microphone;
• and
• receiving at the hardware processor the first audio signal from the user microphone and the second audio signal from the environmental microphone;
• determining via the hardware processor that the first audio signal and the second audio signal both represent the common, distinctive environmental sound emanating from the single environmental source;
• determining via the hardware processor a time correlation adjustment between the first audio signal and the second audio signal, wherein said time correlation adjustment is sufficient to synchronize in time a later-detected environmental sound which is detected at each of the user microphone and the environmental microphone;
• and
• upon subsequently receiving a third sound signal and a fourth sound signal of a later-detected environmental sound which is detected respectively at each of the user microphone and the plurality of environmental microphones, applying said time correlation adjustment to synchronize in time the third sound signal and the fourth sound signal.

A voice-driven system configured for recognition of human speech by identifying a source of a sound as either a user speech or an environmental sound, comprising:

• a user microphone configured to be worn by a user or collocated with a speech organ of the user, and further configured to receive a first sound comprising vocal input from the user and to send to a coupled hardware processor a first audio signal representative of the first sound;
• the coupled hardware processor;
• a plurality of environmental microphones configured to be positioned remotely from the user and distributed at different locations in an environment occupied by the user, and to detect an environmental sound in the environment occupied by the user, wherein at least one environmental microphone of the plurality of environmental microphones is configured to be positioned at a location such that the detected environmental sound excludes the the first sound;
• and
• a transmitter coupled with each environmental microphone of the plurality of environmental microphones, and configured to transmit to the hardware processor a second audio signal representative of the detected environmental sound;
• wherein the hardware processor of the voice-driven system is further configured to:
• receive the first audio signal from the user microphone and the second audio signal from the plurality of environmental microphones;
• determine that the first audio signal and the second audio signal are both indicative of the same one or more simultaneous sound events in the environment;
• determine a relative sound content of the first audio signal and the second audio signal;
• and
• based on the relative sound content, determine that the first audio signal is suitable for identification of words from the user or that the first audio signal is not suitable for identification of words from the user.

A method to recognize human speech by identifying a source of a sound as either a user speech or an environmental sound, the method comprising:

• at a user microphone of a voice-driven system, said user microphone configured to be worn by a user or collocated with a speech organ of the user, receiving a first sound;
• sending from the user microphone to a coupled hardware processor of the voice-driven system a first audio signal representative of the first sound;
• at a plurality of environmental microphones of the voice-driven system configured to be positioned remotely from the user and distributed at different locations in an environment occupied by the user, detecting a second sound in the environment occupied by the user, wherein at least one environmental microphone of the plurality of environmental microphones is configured to be positioned at a location such that the detected second sound excludes the first sound;
• via a transmitter coupled with each environmental microphone of the plurality of environmental microphones, transmitting to the hardware processor a second audio signal representative of the second sound;
• determining via the hardware processor that the first audio signal and the second audio signal are indicative of the same one or more substantially simultaneous sound events in the environment;
• determining via the hardware processor a relative sound content of the first audio signal and the second audio signal;
• and
• based on the relative sound content, determining that the first audio signal is suitable for identification of words from the user or that the first audio signal is not suitable for identification of words from the user.