Martin Swanholm
CTO
Hindenburg Systems
About Me
Martin Swanholm is an experienced software developer, DSP engineer and algorithm designer with a pragmatic and practical perspective, currently specializing in applying machine learning principles to the audio domain. Ongoing work includes various tools and techniques for speech restoration, including phase-coherent frequency-domain convolutional models and multi-task learning models for interactive signal restoration and enhancement workflows.
Sessions
-
Knee-Deep Learning
Practical Steps to Get Started with Audio ML15:00 - 15:50 UTC | Tuesday 12th November 2024 | EmpireBeginnerDive in and start creating! The purpose of this talk is to give a very practical introduction to audio machine learning to audio developers without prior machine learning experience, allowing them to get started quickly on making their own experiments and realizing innovative ideas. Throughout, a simple example model architecture suitable for beginners is used. We dive right in using simple and free tools to acquire data, set up code to create an ML training and inference pipeline, look at training techniques and analyze and evaluate the results as we go. We go through what hardware is needed for training […]