Systems and methods for generating wake signals from known users
Provided herein is an integrated circuit including, in some embodiments, a host processor, a digitally implemented neural network co-processor, and a communications interface between the host processor and the co-processor configured to transmit information therebetween. The special-purpose host processor can be operable as a stand-alone processor. The neural network co-processor may include a digitally implemented neural network. The co-processor is configured to enhance special-purpose processing of the host processor through an artificial neural network. In such embodiments, the host processor is wake keyword identifier processor configured to transmit one or more detected patterns to the co-processor over a communications interface. The co-processor is configured to transmit the recognized patterns to the host processor which can then identify and verify wake keywords spoken by a known user.
1. An integrated circuit, comprising:
a co-processor configured to remain in a low-power, always-on state, the co-processor including an artificial neural network configured to:
generate a wake signal and transmit the wake signal to a host processor; and
identify one or more target signals among one or more signals;
the host processor comprising a keyword identifier host processor configured to receive weighting signals;
wherein synaptic weights of the artificial neural network are; i) modified dynamically based on feedback; ii) accessible in absence of a connection to a communications network; and iii) accessed by a signal detector to differentiate between signals received by the signal detector in absence of the connection to the communications network;
a communications interface between the host processor and the co-processor configured to transmit information therebetween; and
wherein the integrated circuit is configured to operate without an active connection to the communications network.
2. The integrated circuit of claim 1 , wherein the target signal includes at least one wake keyword.
3. The integrated circuit of claim 1 , further comprising verification of a source that includes identifying a speaker of the at least one wake keyword.
4. The integrated circuit of claim 3 , wherein the verification further includes determining if the identified speaker is on a list of pre-authorized users.
5. The integrated circuit of claim 4 , wherein the verification further includes generating verification data in response to verification of the identified speaker on the list of pre-authorized users.
6. The integrated circuit of claim 5 , wherein the co-processor is further configured to transmit verification data to the host processor via the communications interface.
7. The integrated circuit of claim 1 , wherein the wake signal is configured to change a computing device from a first power consumption mode into a second, higher power consumption mode.
8. The integrated circuit of claim 1 , wherein the wake signal is configured to establish a connection between a communication agent and a computing device.
9. The integrated circuit of claim 1 , wherein the signal stream is comprised of signals received via at least one microphone.
10. A method for generating a weight file that causes an integrated circuit to generate wake signals by detecting desired user-specified signals and speakers, comprising:
listing desired target signals and known users that may be detected by a signal detector;
the signal detector configured to:
remain in a low-power, always-on state wherein the signal detector remains continuously ready to detect target information comprising a specific keyword or a known user;
operate in absence of an active connection to a communications network;
generate a wake signal and transmit the wake signal to a host processor; and
retrieving one or more signal databases that are comprised of standard target signals that may be detected by the signal detector;
combining the desired target signals and the one or more signal databases to build a modified database;
using the modified database to train a neural network implementation to recognize the target signals and the standard signals;
producing a set of weights by way of training the neural network implementation;
wherein the weights are i) modified dynamically based on feedback; ii) accessible in absence of a connection to a communications network; and iii) accessed by the signal detector to differentiate between signals received by the signal detector in absence of the connection to the communications network; and
translating the set of weights into the weight file suitable for being stored in a memory storage that is accessible to the integrated circuit; and
wherein the host processor includes a keyword identifier host processor configured to receive weighting signals and transmit one or more signals indicating detection of desired user-specified signals.
11. The method of claim 10 , wherein the target signal includes at least one wake keyword.
12. The method of claim 10 , wherein listing comprises entering the target signals into a computing device that is configured to generate the weight file.
13. The method of claim 10 , wherein listing comprises entering the target signals into a cloud-based application that is configured to generate the weight file.
14. The method of claim 10 , wherein listing comprises entering the target signals into a stand-alone software that is configured to generate the weight file.
15. The method of claim 10 , wherein the target signals are comprised of signal patterns within input signals received via at least one microphone.
16. The method of claim 10 , wherein the target signals may be spoken keywords, or non-verbal acoustic signals such as specific sounds.
17. The method of claim 10 , wherein the neural network implementation is a software model of a neural network that is implemented in the integrated circuit comprising the signal detector.
18. The method of claim 10 , wherein the weight file may be provided to an end-user upon purchasing a mobile device.
19. The method of claim 10 , wherein upon an end-user installing the weight file the mobile device, the signal detector may detect the target signals by way of the set of weights.