Evaluating Semantic Fragility in Text-to-Audio Generation Systems Under Controlled Prompt Perturbations

• URL: http://arxiv.org/abs/2603.13824v1
• Date: Sat, 14 Mar 2026 08:12:40 GMT
• Title: Evaluating Semantic Fragility in Text-to-Audio Generation Systems Under Controlled Prompt Perturbations
• Abstract summary: Small linguistic changes may lead to substantial variation in generated audio, raising concerns about reliability in practical use.<n>We evaluate the semantic fragility of text-to-audio systems under controlled prompt perturbations.
• Score: 2.2870073664564115
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Recent advances in text-to-audio generation enable models to translate natural-language descriptions into diverse musical output. However, the robustness of these systems under semantically equivalent prompt variations remains largely unexplored. Small linguistic changes may lead to substantial variation in generated audio, raising concerns about reliability in practical use. In this study, we evaluate the semantic fragility of text-to-audio systems under controlled prompt perturbations. We selected MusicGen-small, MusicGen-large, and Stable Audio 2.5 as representative models, and we evaluated them under Minimal Lexical Substitution (MLS), Intensity Shifts (IS), and Structural Rephrasing (SR). The proposed dataset contains 75 prompt groups designed to preserve semantic intent while introducing localized linguistic variation. Generated outputs are compared through complementary spectral, temporal, and semantic similarity measures, enabling robustness analysis across multiple representational levels. Experimental results show that larger models achieve improved semantic consistency, with MusicGen-large reaching cosine similarities of 0.77 under MLS and 0.82 under IS. However, acoustic and temporal analyses reveal persistent divergence across all models, even when embedding similarity remains high. These findings indicate that fragility arises primarily during semantic-to-acoustic realization rather than multi-modal embedding alignment. Our study introduces a controlled framework for evaluating robustness in text-to-audio generation and highlights the need for multi-level stability assessment in generative audio systems.

Related papers

• scTranslation: A Comprehensive Benchmark for Single-Cell Multi-Omics Modality Translation [61.97613474494102]
  scTranslation is a comprehensive benchmark for single-cell multi-omics modality translation tasks.<n>It includes diverse translation datasets, integrates state-of-the-art models, and provides a comprehensive evaluation metrics.
  arXiv  Detail & Related papers  (2026-06-02T17:00:49Z)
• UniSonate: A Unified Model for Speech, Music, and Sound Effect Generation with Text Instructions [55.622295453533475]
  We introduce UniSonate, a unified flow-matching framework capable of synthesizing speech, music, and sound effects.<n>We propose a novel dynamic token injection mechanism that projects unstructured environmental sounds into a structured temporal latent space.<n>Experiments demonstrate that UniSonate achieves state-of-the-art performance in instruction-based TTS and TTM.
  arXiv  Detail & Related papers  (2026-04-24T04:26:04Z)
• DEAF: A Benchmark for Diagnostic Evaluation of Acoustic Faithfulness in Audio Language Models [15.637236155997]
  Recent Audio Multimodal Large Language Models (Audio MLLMs) demonstrate impressive performance on speech benchmarks.<n>It remains unclear whether these models genuinely process acoustic signals or rely on text-based semantic inference.<n>We introduce DEAF, a benchmark of over 2,700 conflict stimuli spanning three acoustic dimensions.
  arXiv  Detail & Related papers  (2026-03-17T15:52:26Z)
• Anatomy of the Modality Gap: Dissecting the Internal States of End-to-End Speech LLMs [15.914430317382077]
  We analyze how speech and text representations evolve layer-by-layer.<n>We find that speech representations exhibit a broad cross-layer alignment band, attributable to the redundant nature of speech.
  arXiv  Detail & Related papers  (2026-03-02T06:21:43Z)
• Evaluation of Audio Language Models for Fairness, Safety, and Security [10.580283172294733]
  We introduce a structural taxonomy (system-level and representational) of audio large language models (ALLMs)<n>We propose a unified evaluation framework that assesses semantic invariance under paralinguistic variation, refusal and toxicity behavior under unsafe prompts, and robustness to adversarial audio perturbations.<n>Our findings demonstrate that FSS behavior is tightly coupled to how acoustic information is integrated into semantic reasoning, underscoring the need for structure-aware evaluation of audio language models.
  arXiv  Detail & Related papers  (2026-02-25T21:22:28Z)
• Dynamical Systems Analysis Reveals Functional Regimes in Large Language Models [0.8694591156258423]
  Large language models perform text generation through high-dimensional internal dynamics.<n>Most interpretability approaches emphasise static representations or causal interventions, leaving temporal structure largely unexplored.<n>We discuss a composite dynamical metric, computed from activation time-series during autoregressive generation.
  arXiv  Detail & Related papers  (2026-01-11T21:57:52Z)
• High-Fidelity Speech Enhancement via Discrete Audio Tokens [35.61634772862795]
  DAC-SE1 is a language model-based SE framework leveraging discrete high-resolution audio representations.<n>Our experiments show that DAC-SE1 surpasses state-of-the-art autoregressive SE methods on both objective perceptual metrics and in a MUSHRA human evaluation.
  arXiv  Detail & Related papers  (2025-10-02T16:38:05Z)
• SVeritas: Benchmark for Robust Speaker Verification under Diverse Conditions [54.34001921326444]
  Speaker verification (SV) models are increasingly integrated into security, personalization, and access control systems.<n>Existing benchmarks evaluate only subsets of these conditions, missing others entirely.<n>We introduce SVeritas, a comprehensive Speaker Verification tasks benchmark suite, assessing SV systems under stressors like recording duration, spontaneity, content, noise, microphone distance, reverberation, channel mismatches, audio bandwidth, codecs, speaker age, and susceptibility to spoofing and adversarial attacks.
  arXiv  Detail & Related papers  (2025-09-21T14:11:16Z)
• Where are we in audio deepfake detection? A systematic analysis over generative and detection models [59.09338266364506]
  SONAR is a synthetic AI-Audio Detection Framework and Benchmark.<n>It provides a comprehensive evaluation for distinguishing cutting-edge AI-synthesized auditory content.<n>It is the first framework to uniformly benchmark AI-audio detection across both traditional and foundation model-based detection systems.
  arXiv  Detail & Related papers  (2024-10-06T01:03:42Z)
• High-Fidelity Speech Synthesis with Minimal Supervision: All Using Diffusion Models [56.00939852727501]
  Minimally-supervised speech synthesis decouples TTS by combining two types of discrete speech representations. Non-autoregressive framework enhances controllability, and duration diffusion model enables diversified prosodic expression.
  arXiv  Detail & Related papers  (2023-09-27T09:27:03Z)
• TranSpeech: Speech-to-Speech Translation With Bilateral Perturbation [61.564874831498145]
  TranSpeech is a speech-to-speech translation model with bilateral perturbation. We establish a non-autoregressive S2ST technique, which repeatedly masks and predicts unit choices. TranSpeech shows a significant improvement in inference latency, enabling speedup up to 21.4x than autoregressive technique.
  arXiv  Detail & Related papers  (2022-05-25T06:34:14Z)
• Bridging the Gap Between Clean Data Training and Real-World Inference for Spoken Language Understanding [76.89426311082927]
  Existing models are trained on clean data, which causes a textitgap between clean data training and real-world inference. We propose a method from the perspective of domain adaptation, by which both high- and low-quality samples are embedding into similar vector space. Experiments on the widely-used dataset, Snips, and large scale in-house dataset (10 million training examples) demonstrate that this method not only outperforms the baseline models on real-world (noisy) corpus but also enhances the robustness, that is, it produces high-quality results under a noisy environment.
  arXiv  Detail & Related papers  (2021-04-13T17:54:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.