Designing Smart Algorithms: Traditional DSP vs. Machine Learning

Explore the comparison between traditional DSP and machine learning approaches for designing smart algorithms in this conference talk from ADCx SF. Delve into the world of wakeword detection, tempo detection, and song recognition as Amit Shoham, a senior systems architect and algorithms guru, shares insights on choosing the right mix of techniques for various applications. Gain understanding of fundamental principles that guide algorithm design, considering factors such as system resources and ease of deployment. Learn about the importance of domain knowledge, data sets, and scoring methods in both traditional and machine learning approaches. Discover how to balance the strengths of each method to create efficient and effective smart algorithms for audio applications.