Diffused Responsibility: Analyzing the Energy Consumption of Generative Text-to-Audio Diffusion Models

TL;DR

This paper analyzes the energy consumption of state-of-the-art text-to-audio diffusion models, exploring how generation parameters influence energy use and identifying optimal trade-offs between audio quality and environmental impact.

Contribution

It provides the first comprehensive analysis of energy usage in text-to-audio diffusion models and proposes Pareto-optimal solutions balancing quality and efficiency.

Findings

Energy consumption varies significantly with generation parameters.

Optimal trade-offs exist between audio quality and energy efficiency.

Insights guide development of more sustainable generative audio models.

Abstract

Text-to-audio models have recently emerged as a powerful technology for generating sound from textual descriptions. However, their high computational demands raise concerns about energy consumption and environmental impact. In this paper, we conduct an analysis of the energy usage of 7 state-of-the-art text-to-audio diffusion-based generative models, evaluating to what extent variations in generation parameters affect energy consumption at inference time. We also aim to identify an optimal balance between audio quality and energy consumption by considering Pareto-optimal solutions across all selected models. Our findings provide insights into the trade-offs between performance and environmental impact, contributing to the development of more efficient generative audio models.