System and method for processing speech to identify keywords or other information

A method for perform speech processing comprising steps of:

• a) receiving a signal including speech;
• b) accessing a database of keyword models forming an ensemble of filters associated with each keyword in the database;
• c) decomposing, with an electronic system, the signal including speech into a sparse set of phonetic impulses;
• d) accessing, with an electronic system, the database of keywords and convolve the sparse set of phonetic impulses with the ensemble of filters;
• e) identifying, with an electronic system, keywords within the signal including speech based on step d);
• and
• f) controlling operation the electronic system based on the keywords identified in step e).

An electronic system configured to perform speech processing comprising:

• an audio detection system configured to receive a signal and generate a representation of speech as a sparse set of phonetic events in time;
• a memory having stored therein a database of keyword models forming an ensemble of filters based upon Poisson rate parameters describing an evolution of phonetic events associated with each keyword in the database of keywords;
• a processor configured to:
• i) receive the representation of speech from the audio detection system;
• ii) decompose the representation of speech into a sparse set spare set of phonetic impulses;
• iii) access the database of keywords and convolve the sparse set of phonetic impulses with the ensemble of filters;
• iv) identify keywords within the speech signal based on iii);
• and
• v) control operation the electronic system based on the keywords identified in iv).