MIT's Speech Communication Group (SCG) is developing a system to label landmarks (speech events) in speech data. This tool is expected to be relevant for speech researchers such as linguists, engineers, and speech language pathologists. As it stands, the system is implemented in MATLAB. This poses a hurdle to researchers who may not have technical background with MATLAB. Thus, there is a need to make the system easily available and user-friendly as it continues to grow. My project outlines and implements a solution that may be used to distribute the landmark detection technology out of SCG across research and SLP communities while it grows, weighing privacy, ease-of-use, system performance, and cost.
Praat is a speech analysis software that is very popular amongst speech researchers. Praat is freely available on many platforms and has lots of existing tools and support. For this reason, MATLAB Compiler and UI toolboxes as well as Praat scripting are employed to package the existing SCG system as a Praat plugin. This will make the system available to researchers familiar with Praat. Some testing plans include parallel versus sequential speed comparison, user surveys, and program end-to-end installation on three platforms. Additional materials such as documentation and video demos will be created for the SCG so that the system can be incorporated into their future work.