Explore ultra-low-power command recognition for ubiquitous devices in this 31-minute conference talk from the tinyML Summit 2019. Delve into the challenges of noisy speech recognition, system goals for command recognition, and core functions. Learn about training processes for commands, examine recognition results, and discover example command sets. Investigate implementations on tiny hardware and low-power solutions. Gain insights into the implications of this technology for the future of ubiquitous devices.
Ultra-Low-Power Command Recognition for Ubiquitous Devices
Overview
Syllabus
Intro
The Noisy Speech Problem
Recognition with Noise
Command Recognition System Goals
The Core Functions
Training for Commands
Command Recognition Results
Example Command Set
Implementation on Tiny Hardware
Low power implementations
Implications
Taught by
tinyML
Related Courses
-
Tiny but Powerful: Hardware for High Performance, Low Power Machine Learning
-
tinyMLPerf: Benchmarking Ultra-low Power Machine Learning Systems
-
TinyML Talks France - State of the TinyML Today
-
U-TOE: Universal TinyML On-board Evaluation Toolkit for Low-Power AIoT
-
TinyML Talks - Speech-to-Intent Model Deployment to Low-Power Low-Footprint Devices
-
Dissecting a Low Power AI/ML Edge Application - Noise Suppression
