Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore the fundamentals of machine learning for audio development in this comprehensive conference talk from ADC 2024. Learn practical steps to begin your journey into audio machine learning without requiring prior ML experience, as software developer and DSP engineer Martin Swanholm demystifies theory and implementation complexities through hands-on approaches. Discover how to acquire data using simple and free tools, set up ML training and inference pipelines, explore various training techniques, and analyze results effectively. Understand hardware requirements for training at different scales, from consumer GPUs to cloud computing solutions, while covering basic theory and the historical evolution of different ML approaches. Master data requirements, acquisition methods, and training processes through beginner-friendly model architectures and practical examples. Examine deployment options including cloud-based inference, on-device native code implementation using popular inference frameworks, and dedicated embedded hardware modules for real-time audio processing. Gain insights into solving complex audio problems where traditional DSP methods fall short, whether for audio restoration, speech enhancement, or creative sound generation, with emphasis on pragmatic solutions that work efficiently across various hardware configurations and skill levels.
