NeuraSpeech

NeuraSpeech is a Python toolbox for analyzing EEG responses during continuous speech listening. It includes utilities for phoneme-related potential (PRP) analysis and temporal response function (TRF) modeling.

The package is currently private and under development. Please contact Zhe-chen Guo (zcguo@northwestern.edu) for access or quesitons.

Features

• PRP dataset handling
• Topographic and time-course visualization
• Phoneme manner separability analyses (F-statistic, machine-learning classification, etc.)
• TRF estimation with support for parallel processing

What's new?

09/04/2026: Added an introduction to TRF.

07/12/2026: [Version 0.2.9] Added make_predictors_from_tables() to trf module to allow generation of impulse-like predictors using tables or spreadsheets. See the Predictors from tables tab for usage.

07/10/2026: [Version 0.2.8] Added make_pitch_predictors() to trf module to extract pitch contours and generate pitch predictors. This new functionality is implemented using praat-parselmouth so you would need to install it or update your neuraspeech environment using the updated environment.yml. See the Pitch predictors tab for usage.

See more in Update history.

Quick example

import neuraspeech as ns

# Load data
path_to_data = 'mydata'
path_to_montage = 'my_montage.locs'
prps = ns.PRPData(path_to_data, montage_path = path_to_montage)

ns.plot_manner_prps(
    split_by = ['group'], line_by = 'manner', 
    equal_y_scale = True, lw = 3.0)