Jon Nordby - Sound Event Detection with Machine Learning

Detect specific sound events like bubbles per minute in breweries using machine learning. Learn about the challenges, considerations, and applications of sound event detection, and explore the open-source tool Audacity and its potential applications.

Key takeaways

The talk focuses on sound event detection using machine learning, with a specific application in brew monitoring and fermentation tracking.
The goal is to detect specific events such as bubbles per minute (BPM) in a brewery, which is a challenging task due to variability and noise.
The audience applauded the talk, showing enthusiasm and interest in the topic.
The speaker emphasized the importance of understanding sound detection, event detection, and machine learning.
He introduced the tool Audacity as a great tool for audio processing and feature extraction.
The talk highlighted the challenges and considerations in designing a sound event detection system, including dataset quality, data preprocessing, and model evaluation.
The speaker mentioned that sound event detection is a growing field, with many potential applications in areas such as audio classification, anomaly detection, and IoT.
He emphasized the importance of collaboration and community engagement in advancing the field, mentioning the Sound Sensing community and the Slack group.
The talk concluded with a demonstration of the tool and its potential applications in brew monitoring.
The speaker encouraged attendees to join the Sound Sensing community and try out the tool for themselves.
He also mentioned that the tool is open-source and available on GitHub, and that users can contribute to the project and share their experiences.

Jon Nordby - Sound Event Detection with Machine Learning

More talks