Papers must be formatted according to the instructions in the ICASSP 2026 Paper Submission Guidelines. Also, please take care to read the Conference Policies from the IEEE Signal Processing Society!
If this is your first time to submit a paper to ICASSP, please read the submission guidelines carefully to verify that your paper document is formatted correctly and that you have all the information you need before starting your paper submission. The guidelines contain detailed instructions on formatting your document and completing the submission process, as well as a description of how the review process works and how to prepare for your presentation at the conference if your paper is accepted.
If you are a returning author, you should review the paper guidelines as some information might have changed. Note that the paper kit contains expanded information about the review process and preparing for a presentation at ICASSP 2026.
It is encouraged to include discussions on how your contributions are related to prior work in the field. It is important to put new work in context, to give credit to foundational work, and to provide details associated with the previous work that has appeared in the literature. It is not required that the discussion of relation to prior work be in a separate section, but the discussion must appear in the paper. An optional 5th page may be included, but that page may contain only references.
Exceptional papers and contributors will be selected and recognized by ICASSP.
If you are submitting to a special session, you will receive separate instructions for submitting your paper. Please do not submit invited papers through the regular submission process unless indicated to do so. The formatting requirements in the paper submission guidelines still apply.
Policy on Using Large Language Models (LLMs) in Manuscripts Submitted for Publication
The following rules complement the IEEE rules listed in:
Author Guidelines for Artificial Intelligence (AI)-Generated Text
When submitting a manuscript, the authors implicitly confirm that they have read, understood, and followed the rules for acceptable use of LLMs. In particular, the authors confirm that any output of these tools used in the manuscript has been thoroughly checked, including careful text editing, verification of audio/visual content, and testing of any code to ensure correctness. The basic principle informing the following acceptable use rules is that authors should take full responsibility and ownership for their research and the content of their submitted manuscript. In particular, it is unacceptable for any section of a manuscript to be entirely produced using an LLM.
1. Acceptable uses:
a. For improving language and clarity during the editing process.
b. For accelerating code development and visualization.
c. Research and ideation (identifying related work, feedback on ideas, etc.).
2. Unacceptable uses:
a. Use an LLM to generate most (or significant components) of a manuscript (as opposed to improving the clarity of author-composed text).
b. Direct use of LLM-generated code without subsequent thorough verification of correctness.
c. Direct use of LLM-generated text without subsequent thorough verification of correctness and accuracy for any section of the manuscript, including an introduction, a related work section, and a summary of prior work (distinct from a related work section).
Participation
All papers should be covered by an author registration, and presented in-person at the conference. Uncovered or no-show papers during the virtual conference poster presentation session will be removed from the conference proceedings.