TLDR: The University of Southern California’s Signal Analysis and Interpretation Laboratory (SAIL) has achieved a remarkable ‘hat trick’ at Interspeech 2025, securing three prestigious research awards. Their accolades include the Best Student Paper Award for an AI model distinguishing accents and first and second place in the Speech Emotion Recognition in Naturalistic Conditions Challenge, highlighting significant advancements in human-centered AI and speech communication.
Rotterdam, The Netherlands – The University of Southern California (USC) has made a significant impact at Interspeech 2025, the annual conference of the International Speech Communication Association, with its Signal Analysis and Interpretation Laboratory (SAIL) earning three major honors. The conference, held from August 17–21 in Rotterdam, The Netherlands, recognized SAIL’s groundbreaking contributions to speech processing and human-centered artificial intelligence.
Among the top accolades, USC SAIL received the coveted Best Student Paper Award. This honor was bestowed upon a paper that introduced an innovative AI model capable of mapping out the distinct vocal tract movements that differentiate British and American accents using audio clips. This research, led by PhD student Kevin Huang, marks his second consecutive Best Paper Award at Interspeech, following his recognition in 2024 for work on articulatory settings for L1 and L2 English speakers. The model holds promising applications in accent adaptation, language learning, and advanced speech synthesis, with potential for broader application across various accents and languages.
Further solidifying its leadership, USC SAIL also clinched both first and second place in the conference’s highly competitive Speech Emotion Recognition in Naturalistic Conditions Challenge. This achievement was driven by breakthroughs in accurately predicting emotions from speech, even when accounting for speaker identity, such as gender, and analyzing speech samples that convey mixed emotions. The lab’s new Speech Emotion Recognition (SER) system demonstrated groundbreaking accuracy, outperforming the first runner-up by 35% in the challenge. Traditionally, emotion recognition in speech has relied on associating specific vocal pitches with emotions; however, SAIL’s work has advanced beyond this to achieve superior results in real-world conditions. PhD students Thanathai Lertpetchpun and Jihwan Lee were instrumental in these award-winning efforts.
Shrikanth Narayanan, University Professor and Niki & C. L. Max Nikias Chair in Engineering, and Director of the Ming Hsieh Institute, leads the USC SAIL Lab. Commenting on the achievements, Narayanan stated, “SAIL focuses on human-centered signal and information processing that address key societal needs. Bridging science and engineering, SAILers pioneer approaches that bridge science and engineering to tackle real-world problems, from understanding human speech and emotion to improving communication technologies.”
Also Read:
- Boston University’s Ajay Joshi Joins $1.5 Million NSF Initiative for Light-Powered AI Chips
- NASCIO Honors State Leaders for Pioneering AI and Digital Transformation in Public Service
The recognition at Interspeech 2025 underscores SAIL’s commitment to integrating signal processing, machine learning, and behavioral AI modeling to advance human-centric communication. With 30 papers published and accepted in top conferences and journals in 2025 alone, the lab continues to be a forefront innovator in the field, developing new tools and influential datasets utilized across academia, industry, and society.
