Archival papers
Research papers presenting original empirical or theoretical results; accepted papers will be published in the workshop proceedings.
NLP/CORE Unranked
IMPACT-SPEECH 2026
IMPACT-SPEECH: Identifying, Measuring, Preventing, and Assessing Consequences of Bias in Speech LLMs
IMPACT-SPEECH 2026 is a full-day workshop co-located with EMNLP 2026 that focuses on identifying, measuring, preventing, and assessing bias in speech-enabled large language models and multimodal speech systems. It brings together researchers and practitioners from NLP, speech processing, machine learning, and social sciences to address fairness, accessibility, and ethical challenges in speech AI.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Non-archival papers
Previously published, work-in-progress, and extended abstracts; not published in proceedings but still assessed for relevance.
Research areas in scope
01
Topics of Interest
Fairness evaluation for speech models, including methods to measure disparities across accents, genders, speaker identities, socioeconomic groups, and other factors.Bias detection and benchmarking datasets for speech systems, including the development of datasets designed to identify and evaluate bias in speech recognition, speech generation, and speech-enabled LLMs.Ethical considerations in speech AI development, including responsible data collection, annotation practices, transparency, accountability, and governance of speech technologies.Evaluation of Speech LLMs for fairness and inclusivity, including methods to assess bias in speech understanding, speech generation, and multimodal speech-language systems.Cross-lingual and multilingual fairness, with a focus on challenges faced by underrepresented languages, dialects, and low-resource speech communities.Bias mitigation strategies for speech systems, including algorithmic approaches, training strategies, and fairness-aware adaptation techniques.Real-world impacts of biased speech technologies, including implications for accessibility, employment, digital inclusion, and human–AI interaction.Responsible deployment and governance of speech technologies, including best practices for monitoring, auditing, and mitigating bias in real-world applications.
Policies worth checking twice
- Submissions must be anonymized for a two-way blind review process.
- Archival submissions may be under concurrent review at other venues, but must be withdrawn from those venues if accepted for IMPACT-SPEECH to be included in proceedings.
- Papers must follow the ACL author guidelines format.
- All submission types (long, short, extended abstract) are welcome at any length tier, with page limits of 8, 4, and 2 pages respectively.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026