In an effort to bridge the emotional gap between human and machine communication, a research team at The University of Texas at El Paso has uncovered a subtlety in speech that could revolutionize the way artificial intelligence (AI) systems interact with us. As reported by UTEP Newsfeed, the study led by Dr. Nigel Ward from UTEP's Department of Computer Science suggests that varying degrees of articulation in speech, a phenomenon known as phonetic reduction, may hold the key to creating AI voices that sound more natural and trustworthy.
Dr. Ward's findings are based on a study that analyzed English and Spanish speakers' tendencies to articulate words less precisely when conveying positive emotions. The team recorded participants uttering phrases in both neutral and positive tones, with independent judges then assessing the clarity of the sounds. The results showed a significant rise in phonetic reduction when speakers adopted a positive tone. English speakers displayed 30% reduced and 9% highly reduced pronunciation, while the figures for Spanish speakers stood at 35% and 4%, respectively. This data potentially paves the way for developing AI systems that mimic this human-like speech pattern, as "People often perceive AI systems as cold and unaware and don't trust them, even if their task performance is superb," Dr. Ward said in a statement obtained by UTEP Newsfeed.
The study, which culminated in a paper titled "Phonetic Reduction is Associated with Positive Assessment and other Pragmatic Functions," sheds light on the potential applications for this research. Real-time speech-to-speech translation systems could benefit greatly by incorporating emotional expressiveness, which would make conversations seem less transactional and more relatable. Javier Vazquez, a graduate student who developed the ReduEst tool for estimating phonetic reduction, told UTEP Newsfeed, "We want to move away from just a transactional translation system, where we just kind of get information across, and more so into a conversational system, where we can express feelings and emotions."
