xkl: A legacy software for detailed acoustic analysis of speech made modern

Abstract

The determination of the fundamental properties of speech relies on a fine and precise estimation of temporal and spectral properties of speech segments. Given the time-varying nature of speech, a digital estimation of its instantaneous spectrum is particularly challenging, and has been the object of investigation throughout the past 50 years. The xkl software, developed in the 80’s by the late Dennis Klatt at MIT, has superior capabilities in addressing the above question. Its use in the past 20 years was, however, limited by a lack of support for modern computing platforms. Revamping it will give access to a powerful tool that will eventually lead to new discoveries.

Keywords

• Speech analysis
• Speech spectral analysis
• Spectral display