User programmable voice command recognition based on sparse features
A low power sound recognition sensor is configured to receive an analog signal that may contain a signature sound. Sparse sound parameter information is extracted from the analog signal. The extracted sparse sound parameter information is processed using a speaker dependent sound signature database stored in the sound recognition sensor to identify sounds or speech contained in the analog signal. The sound signature database may include several user enrollments for a sound command each representing an entire word or multiword phrase. The extracted sparse sound parameter information may be compared to the multiple user enrolled signatures using cosine distance, Euclidean distance, correlation distance, etc., for example.
1. A method for training a speaker dependent sound recognition sensor, the method comprising:
receiving an analog signal that contains a command sound spoken by a user of the sound recognition sensor;
extracting sparse sound parameter information from the analog signal using an analog portion of the sound recognition sensor to form a user dependent sound vector representing an entire word or multiword phrase; and
storing the user dependent sound vector in a sound signature database coupled to the sound recognition sensor, wherein the user dependent sound vector is provided to the sound recognition sensor.
2. The method of claim 1 , wherein the user repeats the command sound at least three times, such that at least three user dependent sound vectors are formed by extracting sparse sound parameter information from the analog signal; and
storing the at least three user dependent sound vectors in the sound signature database, such that the at least three user dependent sound vectors are provided to the sound recognition sensor.
3. The method of claim 1 , wherein the sound signature database comprises a plurality of sound signatures each representing a whole spoken word or multiword phrase.
4. A method for operating a speaker dependent sound recognition sensor, the method comprising:
receiving an analog signal that may contain a trigger sound;
extracting sparse sound parameter information from the analog signal using an analog portion of the sound recognition sensor; and
processing the extracted sparse sound parameter information using a speaker dependent sound signature database stored in the sound recognition sensor to identify sounds or speech contained in the analog signal, wherein the sound signature database comprises a plurality of user dependent sound vectors from a single user each representing a same entire word or multiword phrase.
5. The method of claim 4 , wherein the sound signature database comprises a plurality of sound signatures each representing a whole spoken word or multiword phrase.
6. An apparatus for recognizing a sound, the apparatus comprising:
an analog front end section comprising analog feature extraction circuitry configured to receive an analog signal from a microphone that may contain a signature sound and to extract sparse sound parameter information from the analog signal; and
a digital classification section coupled to the analog front end section and configured to compare the sound parameter information to a sound signature database stored in memory coupled to the digital classification section to detect when the signature sound is received in the analog signal and to generate a match signal when the signature sound is detected, wherein the sound signature database comprises a plurality of speaker dependent sound vectors from a single speaker each representing a same entire word or multiword phrase.