Signal quality assessment is a fundamental problem in signal processing, with broad implications for compression, restoration, generation, and human-computer interaction. As foundation models such as CLIP, GPT-4, and LLaVA revolutionize cross-modal understanding and signal generation, they also introduce new challenges and opportunities for evaluating the quality of signals, whether visual, acoustic, textual, or multimodal. Therefore, to explore signal quality assessment from two perspectives is significant: utilizing foundation models as assessors of signal quality, and evaluating the intrinsic quality or alignment performance of foundation models themselves.

This special session will cover signal quality assessment research involving foundation models across modalities. Topics of interest include, but are not limited to:

• Signal quality assessment using foundation models
• Evaluation of foundation models, including alignment, fidelity, and robustness
• Zero-shot or few-shot quality assessment via foundation models
• Cross-modal signal consistency and semantic quality assessment using foundation models
• Quality-aware fine-tuning or adaptation of foundation models
• Dataset design and benchmark for foundation model-driven quality assessment
• Interpretability and explainability in signal quality assessment using foundation models

Submission website: https://cmsworkshops.com/ICASSP2026/papers/submission.asp?SessionType=Special

After signing in the ICASSP 2026 submission site as the author, please choose our Special Session name to submit the paper. Papers must be no longer than 5 pages, including all text, figures, and 1-page references. ICASSP 2026 reviewing is single blind, which means that authors cannot know the names of the reviewers of their papers, and reviewers can know the names of the authors.

• Paper submission deadline:    17th September 2025