Photo by Tony RinaldoPhoto by Tony Rinaldo
Maria-GabriellaDi Benedetto
2019–2020
William Bentinck-Smith Fellow
Sapienza University of Rome (Italy)
Engineering Sciences
Toward Speech Recognition of the Italian Language Based on Detection of Landmarks and Other Acoustic Cues to Features

Maria-Gabriella Di Benedetto is a professor of communication theory and engineering at Sapienza Università di Roma, in Italy. Her major areas of research are speech and wireless communications, with attention to ultra-wideband impulse radio (IR-UWB). The advances she made in understanding complex IR-UWB signals recently focused her research interest on integrating this knowledge with her long-term expertise in speech, a signal that is intrinsically IR-UWB.

Creating an automatic speech recognizer that imitates the process through which a listener deciphers the words intended by a speaker requires understanding how lexical items are stored in memory. At Radcliffe, Di Benedetto is investigating the lexical access model proposed by Kenneth N. Stevens for American English and extending it to Italian. She hopes to understand whether Stevens’s approach has universal application across languages, with relevant implications for how the human brain recognizes speech. This investigation will form the foundation for the development of a complete speech recognition system for Italian.

Di Benedetto obtained her PhD in electrical engineering in 1987 from Sapienza Università di Roma. She has been principal investigator of several European Union projects on advanced communication systems design, gathering more than 150 researchers worldwide. She has held visiting positions at the Massachusetts Institute of Technology, the University of California, Berkeley, and Paris-Sud University, in France. She received the John H. MacKay, Jr., Professorship in Electrical Engineering from the University of California, Berkeley, and a Sapienza Ricerca award for excellence in research. She is a Fellow of the IEEE.


2019–2020 Radcliffe Institute Fellows

This information is accurate as of the fellowship year indicated for each fellow.
Photo by Tony Rinaldo