Energy Consumption Trends in Sound Event Detection Systems

TL;DR

This paper analyzes how integrating energy consumption metrics into sound event detection challenges influences system development, highlighting trends towards energy efficiency amidst increasing system complexity.

Contribution

It introduces energy consumption evaluation in SED challenges and examines its impact on system design and performance over three years.

Findings

Shift towards more energy-efficient training methods

Increase in system complexity despite energy focus

Energy metrics influence challenge evaluation criteria

Abstract

Deep learning systems have become increasingly energy- and computation-intensive, raising concerns about their environmental impact. As organizers of the Detection and Classification of Acoustic Scenes and Events (DCASE) challenge, we recognize the importance of addressing this issue. For the past three years, we have integrated energy consumption metrics into the evaluation of sound event detection (SED) systems. In this paper, we analyze the impact of this energy criterion on the challenge results and explore the evolution of system complexity and energy consumption over the years. We highlight a shift towards more energy-efficient approaches during training without compromising performance, while the number of operations and system complexity continue to grow. Through this analysis, we hope to promote more environmentally friendly practices within the SED community.