Meta Audiobox Aesthetics: Unified Automatic Assessment for Speech, Music and Sound
Authors:
- 6 December 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Abstract
Quantifying audio aesthetics is challenging due to its subjective nature, influenced by human perception and cultural context. Traditional methods rely on human listeners, leading to inconsistencies and high resource demands. This paper addresses the growing need for automated systems capable of predicting audio aesthetics without human intervention. Such systems are crucial for applications like data filtering, pseudo-labeling, and evaluating generative models.In this paper, we propose new annotation guidelines that break down human listening perspectives into four axes and develop no-reference, peritem prediction models for more nuanced audio quality assessment. Our models are evaluated against human mean opinion scores (MOS) and existing methods, demonstrating comparable or superior performance. This research not only advances the field of audio aesthetics but also provides open-source models and datasets to facilitate future work and benchmarking.Keywords
All Related Versions
This publication has 27 references indexed in Scilit:
- LE-SSL-MOS: Self-Supervised Learning MOS Prediction with Listener Enhancement2023
- Large-Scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation2023
- Torchaudio-Squim: Reference-Less Speech Quality and Intelligibility Measures in Torchaudio2023
- WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech ProcessingIEEE Journal of Selected Topics in Signal Processing, 2022
- Generalization Ability of MOS Prediction Networks2022
- Dnsmos: A Non-Intrusive Perceptual Objective Speech Quality Metric to Evaluate Noise Suppressors2021
- ViSQOL v3: An Open Source Production Ready Objective Speech and Audio Metric2020
- Espnet-TTS: Unified, Reproducible, and Integratable Open Source End-to-End Text-to-Speech Toolkit2020
- Audio Set: An ontology and human-labeled dataset for audio events2017
- Perceptual evaluation of speech quality (PESQ)-a new method for speech quality assessment of telephone networks and codecs2002