Interface to leaky spiking neurons
A processor, that may include at least one neural network that comprises at least one leaky spiking neuron; wherein the at least one leaky spiking neuron is configured to directly receive an input pulse density modulation (PDM) signal from a sensor; wherein the input PDM signal represents a detected signal that was detected by the sensor; and wherein the at least one neural network is configured to process the input PDM signal to provide an indication about the detected input signal.
1. A processor, that comprises:
at least one neural network that comprises at least one leaky spiking neuron; wherein the at least one leaky spiking neuron is configured to directly receive an input pulse density modulation (PDM) signal from a sensor; wherein the input PDM signal represents a detected signal that was detected by the sensor; and
wherein the at least one neural network is configured to process the input PDM signal to provide an indication about the detected input signal.
2. The processor according to claim 1 wherein the sensor is an audio sensor and the detected signal is an audio signal.
3. The processor according to claim 2 wherein the at least one neural network comprises a first group of leaky spiking neurons that is configured to pre-process the PDM signal to provide approximations of output signals of a bank of Mel filters, the approximations represent the detected signal.
4. The processor according to claim 3 wherein the at least one neural network comprises a second group of leaky spiking neurons that is coupled to the first group of leaky spiking neurons and is configured to process the approximations to provide an audio process result.
5. The processor according to claim 3 wherein the first group of leaky spiking neurons comprises multiple resonators, wherein different resonators are configured to output approximations of different Mel filters of the bank of MEL filters.
6. The processor according to claim 5 wherein the multiple resonators are multiple tunable resonators.
7. The processor according to claim 5 wherein each resonator of the multiple resonators comprises a loop of leaky spiking neurons that is followed by a peak detector.
8. The processor according to claim 5 wherein the peak detector comprises a first leaky spiking neuron that functions as a diode and a a second leaky spiking neuron that smoothes an output signal of the first leaky spiking neuron.
9. The processor according to claim 3 wherein the at least one neural network is configured to process the PDM signal to provide an audio process result without calculating Mel-frequency cepstral coefficients.
10. The processor according to claim 1 wherein the detected signal differs from an audio signal.
11. The processor according to claim 1 comprising an interface driver that is located between the sensor and the at least one leaky spiking neuron that is configured to receive the input PDM signal, wherein the interface driver is configured to match between an input impedance of the at least one neural network and an output impedance of the sensor.
12. A method for signal-processing, the method comprises:
directly receiving, by at least one leaky spiking neuron of at least one neural network directly, an input pulse density modulation (PDM) signal from a sensor; wherein the input PDM signal represents a detected signal that was detected by the sensor; and
processing, by the at least one neural network, the input PDM signal to provide an indication about the detected input signal.
13. The method according to claim 12 wherein the sensor is an audio sensor and the detected signal is an audio signal.
14. The method according to claim 13 wherein the at least one neural network comprises a first group of leaky spiking neurons wherein the method comprises pre-processing, by the first group of leaky spiking neurons, the PDM signal to provide approximations of output signals of a bank of Mel filters, the approximations represent the detected signal.
15. The method according to claim 14 wherein the at least one neural network comprises a second group of leaky spiking neurons that is coupled to the first group of leaky spiking neurons, and wherein the method comprises processing, by the second group of leaky spiking neurons, the approximations to provide an audio process result.
16. The method according to claim 14 wherein the first group of leaky spiking neurons comprises multiple resonators, wherein the method comprises outputting, by different resonators, approximations of different Mel filters of the bank of MEL filters.
17. The method according to claim 16 wherein the multiple resonators are multiple tunable resonators.
18. The method according to claim 16 wherein each resonator of the multiple resonators comprises a loop of leaky spiking neurons that is followed by a peak detector.
19. The method according to claim 16 wherein the method comprises truncating to zero, by a first leaky spiking neuron of a peak detector, negative output signals of the loop; and smoothing, by a second leaky spiking neuron of the peak detector, an output signal of the first leaky spiking neuron of the peak detector.
20. The method according to claim 14 comprising processing the PDM signal to provide an audio process result without calculating Mel-frequency cepstral coefficients.
21. The method according to claim 12 wherein the detected signal differs from an audio signal.
22. The method according to claim 12 comprising an interface driver that is located between the sensor and the at least one leaky spiking neuron; wherein the method comprises matching, by the interface driver, between an input impedance of the at least one neural network and an output impedance of the sensor.