Demonstrating Parselmouth: Integrating Praat into a complex Python workflow

Interdisciplinary science requires interdisciplinary tools: as different scientific disciplines combine methods and approaches to research language and speech, existing algorithms and tools need to be combined. One of these tools is Praat (Boersma and Weenink, 2018), a software package implementing a wide range of acoustic and phonetic algorithms and analyses. While Python and other scripting languages are designed such that they allow automation and gluing together of different parts of the research into a single workflow, doing so with Praat and its scripting language is not always as straightforward. To simplify the integration of Praat into a complex workflow, we have developed Parselmouth.
Parselmouth is an open-source Python library 1,2 that allows one to access Praat functionality and combine it with other Python tools and libraries, yet feels natural and simple to a Python user. As such, we believe Parselmouth can be useful in a variety of cases, ranging from the visualization of acoustic data or batch analysis of audio files, to computational models on speech perception and acquisition.
We also introduce the possibility of integrating Praat’s phonetic analysis and manipulation of responses and stimuli in an adaptive experiment, using Parselmouth. One widely-used Python framework for setting up and running "a wide range of neuroscience, psychology and psychophysics experiments" is PsychoPy (Peirce, 2007), which includes a graphical interface to build experiments with a minimal amount of coding.
We demonstrate the integration of Parselmouth into PsychoPy experiments with an adaptive staircase experiment (e.g., Kaernbach, 2001; de Boer, 2012) to determine the minimal amount of noise that stops participants from correctly classifying a stimulus. Using Praat functionality through Parselmouth, at each step during the experiment a new audio stimulus is created with the desired signal-to-noise ratio. With the presented examples, we hope to illustrate the new, modern experimental setups and workflows that Parselmouth facilitates, advancing interdisciplinary research to answer questions in language sciences.