Fréchet Audio Distance: A Metric for Evaluating Music Enhancement Algorithms

Abstract

We propose the Fréchet Audio Distance (FAD), a novel, reference-free evaluation metric for music enhancement algorithms. We demonstrate how typical evaluation metrics for speech enhancement and blind source separation can fail to accurately measure the perceived effect of a wide variety of distortions. As an alternative, we propose adapting the Fréchet Inception Distance (FID) metric used to evaluate generative image models to the audio domain. FAD is validated using a wide variety of artificial distortions and is compared to the signal based metrics signal to distortion ratio (SDR), cosine distance and magnitude L2 distance. We show that, with a correlation coefficient of 0.52, FAD correlates more closely with human perception than either SDR, cosine distance or magnitude L2 distance, with correlation coefficients of 0.39, -0.15 and -0.01 respectively.

References in corpus (3)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
• YouTube-8M: A Large-Scale Video Classification Benchmark

Cited by in corpus (7)

• Speech Gesture Generation from the Trimodal Context of Text, Audio, and Speaker Identity
• High Fidelity Speech Synthesis with Adversarial Networks
• Taming Visually Guided Sound Generation
• MTCRNN: A multi-scale RNN for directed audio texture synthesis
• DarkGAN: Exploiting Knowledge Distillation for Comprehensible Audio Synthesis with GANs
• Generative Models for Security: Attacks, Defenses, and Opportunities
• Towards Lightweight Controllable Audio Synthesis with Conditional Implicit Neural Representations