EAGER: HNDS-I: Next steps in acoustic phonetics research

Progress in science and technology often depends on foundational software tools that make it possible to find patterns in large data sets and visualize complex processes. This is certainly true in the study of speech communication, which contributes to the development of consumer tools such as Siri and Alexa, and the development of software tools for voice verification. This proposal builds a new visualization application and a software toolkit for teaching and research in speech communication. The activities funded by the proposal fall into two main categories (1) to write and test software for speech analysis, and (2) to get input and guidance for this software from leaders in both academia and industry, ensuring wide utility and adoption. Thus, the project directly advances research in artificial intelligence, particularly the use of AI techniques in speech communication. Other benefits to society include translational research because the software developed in this project is used in speech technology product development and testing, in clinical speech applications for interactive speech assessment and therapy, and in teaching speech communication, phonetics, and language sciences.

This project will develop a more modern software tool to be used in acoustic phonetic research. The first and primary aim of the project develops a graphic user interface and expanded toolkit in the widely adopted and supported Python programming language. This effort is built on an existing software library and leverages modern infrastructure for cross-platform deployment in a stable software environment, which is easily maintainable and documented, and has a wide range of preexisting libraries for signal processing and database management that are available in a standard software development environment. This rich software infrastructure facilitates establishing a wide user group. To ensure that the development effort meets the needs of the research community, the work is done in consultation with an advisory board drawn from both academia and industry. The second aim creates and publishes several demonstration studies, producing original analyses of existing speech corpora that contribute new knowledge and serve as models for other researchers. The third aim establishes an online user community through on-line workshops and organized with the help of advisory committee members and develops online documentation featuring both “how-to” videos and traditional documentation pages for the functions in the toolkit. One anticipated outcome of the project is the establishment of strong links with industry partners.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Award Number: 2538747
Principal Investigator: Keith Johnson
Funds Obligated: $300,000
State: CA