Prosody and natural human speech in machine learning.
Speech and Learning
Courante observes prosodic and phonetic fluency in readers of text and music in order to help companies build ML solutions for natural human speech synthesis and phrasal flow.
Dr. JOCELYN DUECK
Dr. Jocelyn Dueck is Assistant Professor of Vocal Coaching and Collaborative Arts at Carnegie Mellon University and founder of Courante LLC, a company that assists humans and machines in speech. A subject matter expert in prosody, Dr. Dueck’s work with the Adobe team focuses on analyzing voice recordings.
Prosody refers to stress and intonation patterns in language. In essence, it is the music of the spoken word.
Dr. Dueck’s background in classical piano encompasses more than 45 years dedicated to fluency in phrasing while reading music – as collaborative pianists read exceptionally well on sight for hours a day. For 30 of those years, Dr. Dueck has listened to and prepared opera singers as a coach, including work at The Juilliard School, Manhattan School of Music, Bard College, the Metropolitan Opera and Wolf Trap Opera. Dr. Dueck’s techniques include keen observations of singers’ mouths and reading skills, detailed hearing of the acoustic properties of their voices in rooms large and small, as well as traditional musical coaching and teaching languages including French, Italian, German and English and their related dialects and accents.
The prosody of the spoken word functions much like music – speech has volume, pitch, duration and rhythm. Whenever we read or speak, we are composers. The key is discovering how to compose with greater ease and fluency.
Get In Touch
jocelyn [at] courante.com