END-TO-END NEUROMORPHIC ACOUSTIC PROCESSING

Organization Name

Inventor(s)

Katarzyna J. Kaszuba-miotke of Gdynia (PL)

END-TO-END NEUROMORPHIC ACOUSTIC PROCESSING - A simplified explanation of the abstract

This abstract first appeared for US patent application 18148358 titled 'END-TO-END NEUROMORPHIC ACOUSTIC PROCESSING

The abstract describes a patent application for a neuromorphic processing device that utilizes a spike generator to process acoustic signal data and implement a spiking neural network for acoustic recognition tasks.

Simplified Explanation:

- A neuromorphic processing device uses a spike generator to process acoustic signals and implement a spiking neural network for acoustic recognition tasks.

Key Features and Innovation:

- Spike generator generates input spikes based on acoustic signal data. - Neuromorphic compute block implements a spiking neural network (SNN). - SNN receives input spikes and generates output spikes for acoustic recognition tasks.

Potential Applications:

- Speech recognition systems - Voice-controlled devices - Acoustic event detection

Problems Solved:

- Efficient processing of acoustic signals - Real-time acoustic recognition tasks

Benefits:

- Improved accuracy in acoustic recognition - Faster processing of acoustic data - Reduced power consumption compared to traditional methods

Commercial Applications:

- Potential use in smart speakers, virtual assistants, and IoT devices for enhanced acoustic recognition capabilities.

Questions about Neuromorphic Processing Device:

1. How does the spike generator process acoustic signal data? - The spike generator uses hardware to generate a set of input spikes based on the acoustic signal data received from a microphone.

2. What is the role of the neuromorphic compute block in the device? - The neuromorphic compute block implements a spiking neural network (SNN) to receive the input spikes and generate output spikes for acoustic recognition tasks.

Original Abstract Submitted

A neuromorphic processing device includes a spike generator including hardware to generate a set of input spikes based on acoustic signal data generated by a microphone of a computing device. The neuromorphic processing device further includes a neuromorphic compute block to implement a spiking neural network (SNN), receive the set of input spikes as an input to the SNN, and generate a set of output spikes from the SNN based on the input. A result for an acoustic recognition task may be determined based on the set of output spikes.