Prototype-Rectified Iterative Self-supervised Manifold Denoising under Severe Acoustic Shift

Hugging Face Daily Papers Papers

Summary

PRISM is a training-free test-time adaptation method that reverses low-rank affine noise distortions in audio-text models using frozen text prototypes, showing significant improvements under severe acoustic noise.

Audio-Text Foundation Models (ATMs) fail catastrophically under severe acoustic noise, yet existing adaptation strategies either rely on gradient-based Test-Time Adaptation (TTA), which reinforces noise rather than signal, or on prompt tuning that requires privileged noise annotations unavailable at inference. We address these failures with PRISM (Prototype-Rectified Iterative Self-supervised Manifold Denoising), a training-free, source-free TTA framework grounded in the Affine Noise Hypothesis: severe acoustic noise induces a low-rank affine shift in the multimodal latent space, with more than 90% of distortion energy confined to the leading 60 principal components. PRISM estimates and reverses this distortion from an unlabeled target batch using frozen text prototypes as geometric anchors via three closed-form geometric corrections compiled into a single static projection matrix by Affine Bias Regression. At inference, adaptation reduces to one matrix-vector multiplication in 0.0009 ms, making it substantially faster than gradient-based TTA while requiring no additional training. On UrbanSound8K, PRISM improves over the zero-shot baseline by 12.94 percentage points and surpasses an oracle-assisted TTA baseline by 9.41 percentage points, despite never observing its privileged augmented noise prompts. We further identify the Polyphonic Trap, a principled failure mode of subspace deflation for broadband classes, and resolve it via Confidence-Aware Regression (CAR), recovering up to 8.16 percentage points for the worst-affected class.
Original Article
View Cached Full Text

Cached at: 08/18/26, 07:50 AM

Paper page - Prototype-Rectified Iterative Self-supervised Manifold Denoising under Severe Acoustic Shift

Source: https://huggingface.co/papers/2608.15037

Abstract

PRISM is a fast, training-free test-time adaptation method that reverses low-rank affine noise distortions in audio-text models using frozen text prototypes and geometric corrections.

Audio-Text Foundation Models(ATMs) fail catastrophically under severe acoustic noise, yet existing adaptation strategies either rely on gradient-basedTest-Time Adaptation(TTA), which reinforces noise rather than signal, or on prompt tuning that requires privileged noise annotations unavailable at inference. We address these failures with PRISM (Prototype-Rectified Iterative Self-supervised Manifold Denoising), a training-free, source-free TTA framework grounded in theAffine Noise Hypothesis: severe acoustic noise induces a low-rank affine shift in the multimodal latent space, with more than 90% of distortion energy confined to the leading 60 principal components. PRISM estimates and reverses this distortion from an unlabeled target batch using frozen text prototypes as geometric anchors via three closed-form geometric corrections compiled into a single static projection matrix byAffine Bias Regression. At inference, adaptation reduces to one matrix-vector multiplication in 0.0009 ms, making it substantially faster than gradient-based TTA while requiring no additional training. On UrbanSound8K, PRISM improves over the zero-shot baseline by 12.94 percentage points and surpasses an oracle-assisted TTA baseline by 9.41 percentage points, despite never observing its privileged augmented noise prompts. We further identify thePolyphonic Trap, a principled failure mode of subspace deflation for broadband classes, and resolve it viaConfidence-Aware Regression(CAR), recovering up to 8.16 percentage points for the worst-affected class.

View arXiv pageView PDFGitHub1Add to collection

Get this paper in your agent:

hf papers read 2608\.15037

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2608.15037 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2608.15037 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2608.15037 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

PRISM: Priority-aware Rubric Internalization via Structured Multimodal Data Synthesis

arXiv cs.LG

This paper introduces PRISM, a four-stage data synthesis framework for training multimodal LLMs to follow prioritized rubrics, and PRISM-Eval, a judge-free evaluation suite. With only 10K samples, PRISM lifts Qwen3-VL-4B from 9.5% to 30.1% Strict accuracy on PRISM-Eval while preserving general benchmark performance, and gains transfer to other open-source MLLMs.