Jump to Main Content

Carnegie Speech

To be understood and to understand - because technology matters


Dr. Jaime Carbonell

Muscle memory is not usually associated with learning a new language. But to Dr. Jaime Carbonell, Director of Carnegie Mellon University Language Technologies Institute (LTI) and Chairman of Carnegie Speech, muscle memory is critical to speaking a new language. Since correct articulatory positioning is necessary for intelligible pronunciation, practicing this positioning until it becomes automatic is critical. Drills and practice really do work as long as they are pinpointed to those skills that need to be improved and include understandable and actionable advice for improvement.

Originally from Montevideo, Uruguay and fluent in multiple languages, Jaime knows from personal experience that incorrectly applied sounds, pace and intonation from the speaker's native language can result in pronunciation errors that are difficult and expensive to remedy.

Taking on this challenge, Jaime and Carnegie Mellon University's Language Technologies Institute, featuring 200 dedicated researchers, explored contrastive language analyses to identify likely and significant pronunciation gaps. Through computer-enabled physical exercises that addressed pronunciation errors, the articulatory patterns required to speak a new language effectively were developed and reinforced.

Theory was nice – the reality was better. Language labs and speech muscle memory exercises were enthusiastically received by students in initial trials. With hypotheses validated, Jaime and Dr. Maxine Eskenazi formed Carnegie Speech to develop spoken language training products that use computer assisted learning tools and speech recognition technologies to identify, pinpoint and correct speech challenges. With the introduction of personalized curriculum, pinpointed language analysis and instant feedback incorporated in an Intelligent Tutor, Carnegie Speech's products have made significant impact with global companies in multiple sectors

Dr. Maxine Eskenazi

It all started with the two Suzannes.

While visiting Paris as a student, Dr. Maxine Eskenazi found her French being corrected and critiqued by two fellow classmates – both named Suzanne. Despite being conversant in French, the constant and negative feedback discouraged Maxine from speaking French on that trip, but did plant the seed for what one day would become Carnegie Speech.

The idea of non-judgmental and self-paced spoken language instruction that provided immediate feedback to the student continued to percolate after graduation, especially as Maxine taught French to Americans and then taught English to French students. Recalling the importance of student motivation and self-confidence necessary to learn a foreign language, Maxine wanted to create a learning technology that enabled private, effective and self-paced spoken language practice.

After years of being a speech recognition researcher in France, Maxine returned to Carnegie Mellon University. During her work at Carnegie Mellon with the Sphinx speech recognition technology, Maxine realized that, with proper engineering, it was now possible to build the language learning system she had been imagining. As a faculty member of Carnegie Mellon University's noted Language Technologies Institute (LTI), Maxine's concept of using computers to teach spoken language in a self-paced, effective non-judgmental manner was finally tested – and was met with immediate success. Participants in initial studies refused to stop using the technology, going so far as to ask lab instructors to let them use it long after the initial studies were completed.

Armed with these personal, academic and technical experiences, Maxine and Director of CMU's Language Technologies Institute (LTI) – Dr. Jaime Carbonell – started a new venture: Carnegie Speech. Now, thanks to the two Suzanne's and Dr. Eskenazi's vision and expertise, many global companies, as well as thousands of end-users, use Carnegie Speech products to improve their spoken language skills.

Vist Dr. Maxine Eskenazi's website for additional project and research information