mlsanityData to 14 Oct 2026

Improving Perceptual Audio Aesthetic Assessment via Triplet Loss and Self-Supervised Embeddings

arXiv 2509.03292 ↗First posted 3 Sep 2025eess.AS cs.LG cs.SDcited by 1

Not on the map. Why?

Abstract (arXiv)

We present a system for automatic multi-axis perceptual quality prediction of generative audio, developed for Track 2 of the AudioMOS Challenge 2025. The task is to predict four Audio Aesthetic Scores--Production Quality, Production Complexity, Content Enjoyment, and Content Usefulness--for audio generated by text-to-speech (TTS), text-to-audio (TTA), and text-to-music (TTM) systems. A main challenge is the domain shift between natural training data and synthetic evaluation data. To address this, we combine BEATs, a pretrained transformer-based audio representation model, with a multi-branch long short-term memory (LSTM) predictor and use a triplet loss with buffer-based sampling to structure the embedding space by perceptual similarity. Our results show that this improves embedding discriminability and generalization, enabling domain-robust audio quality assessment without synthetic training data.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 21 of its 22 references that are in the map. First posted on arXiv 3 Sep 2025.

Builds on

21 of 22 references.

References · 21

  1. 2015FaceNet: A Unified Embedding for Face Recognition and Clusteringcited by 2,849
  2. 2018Unsupervised Feature Learning via Non-Parametric Instance-level Discriminationcited by 1,845
  3. 2022UTMOS: UTokyo-SaruLab System for VoiceMOS Challenge 2022cited by 390
  4. 2022BEATs: Audio Pre-Training with Acoustic Tokenizerscited by 388
  5. 2025Meta Audiobox Aesthetics: Unified Automatic Quality Assessment for Speech, Music, and Soundcited by 137
  6. 2021Generalization Ability of MOS Prediction Networkscited by 69
  7. 2022The VoiceMOS Challenge 2022cited by 57
  8. 2021MBNet: MOS Prediction for Synthesized Speech with Mean-Bias Networkcited by 37
  9. 2021Deep Learning-based Non-Intrusive Multi-Objective Speech Assessment Model with Cross-Domain Featurescited by 33
  10. 2021LDNet: Unified Listener Dependent Modeling in MOS Prediction for Synthetic Speechcited by 24
  11. 2022SpeechLMScore: Evaluating speech generation using speech language modelcited by 20
  12. 2024The VoiceMOS Challenge 2024: Beyond Speech Quality Predictioncited by 17
9 more
  1. 2024Enabling Auditory Large Language Models for Automatic Speech Quality Evaluationcited by 13
  2. 2023A Study on Incorporating Whisper for Robust Speech Assessmentcited by 11
  3. 2022Fusion of Self-supervised Learned Models for MOS Predictioncited by 10
  4. 2024Non-Intrusive Speech Intelligibility Prediction for Hearing-Impaired Users using Intermediate ASR Features and Human Memory Modelscited by 9
  5. 2023The VoiceMOS Challenge 2023: Zero-shot Subjective Speech Quality Prediction for Multiple Domainscited by 8
  6. 2024Self-Supervised Speech Quality Estimation and Enhancement Using Only Clean Speechcited by 7
  7. 2024Speech foundation models on intelligibility prediction for hearing-impaired listenerscited by 6
  8. 2024HAAQI-Net: A Non-intrusive Neural Music Audio Quality Assessment Model for Hearing Aidscited by 3
  9. 2023CORN: Co-Trained Full- And No-Reference Speech Quality Assessmentcited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Robust Generative Audio Quality Assessment: Disentangling Quality from Spurious Correlationscited by 0