06-18, 11:45–12:30 (Europe/London), Tower Suite 1
Training audio neural networks requires creating or using pre-existing manually tagged data. In this talk we will review the state of the art algorithms that automate this process and show how they can help in real-world use-cases.
Manual listening tests are great but they’re time consuming, mission specific and expensive. We all want good quality automated testing measurements to better our algorithms but can we truly get there?
Advanced techniques such as Visqol (Google ,2020) and NORESQA (Facebook 2021) are recent open source tools used to achieve automated model testing. They all aim to close the gap between our audio perception and the raw signal. Using them can help find the right path towards improvement, and have more confidence in our models.
This talk will point you in the right direction to start using automated audio tests in your work. We will explore the field through Python’s Librosa library and go over the fundamental concepts and basic usage.
We’ll also get a feeling for the power and variety of real-world use cases, by creating a test that explores a real-world example finding bugs in edge cases we didn’t know. We’ll finish with pointers and resources to help you get started.
Previous knowledge expected
