Explore the future of AI in audio processing with this tinyML talk by Vikram Shrivastava, Sr. Director of IoT Marketing at Knowles Corporate. Discover how dedicated audio processors at the edge are revolutionizing AI applications, offering improved user interfaces with lower latency and reduced costs. Learn about the challenges of implementing voice integration, essential requirements, and design capabilities needed for effective deployment. Gain insights into control interface integration, software stacks, algorithm development, and user space application development. Delve into topics such as fully connected networks, activation functions, TensorFlow Lite Micro, and Knowles DSPs. Understand the process of formatting coefficients, using the TensorFlow Lite Converter, and implementing audio classifiers. Explore real-world case studies and get answers to common questions about memory, multiple microphones, and package sizes in this comprehensive 52-minute presentation.
Dedicated Audio Processors at the Edge Are the Future of AI
Overview
Syllabus
Intro
tinyML Talk Sponsors
tinyML Vision Challenge
tinyML Tutorials
Vikrams Background
About Knowles
Applications of Knowles DSPs
Knowles DSPs
Fully Connected Networks
Activation Functions
MVM Library
TensorFlow Lite Micro
TensorFlow Lite Micro kernels
Knowles TensorFlow Lite Converter
TensorFlow Lite Micro Interpreter
Formatting coefficients
tflight converter
tensorflow audio classifier
mcps
case study
summary
support
QA
Sphinx
UC Size
QFN Package
Memory
Multiple Mics
Thank You
Taught by
tinyML
