Introducing MEG-MASC a high-quality magneto-encephalography dataset for evaluating natural speech processing

Abstract

The "MEG-MASC" dataset provides a curated set of raw magnetoencephalography (MEG) recordings of 27 English speakers who listened to two hours of naturalistic stories. Each participant performed two identical sessions, involving listening to four fictional stories from the Manually Annotated Sub-Corpus (MASC) intermixed with random word lists and comprehension questions. We time-stamp the onset and offset of each word and phoneme in the metadata of the recording, and organize the dataset according to the 'Brain Imaging Data Structure' (BIDS). This data collection provides a suitable benchmark to large-scale encoding and decoding analyses of temporally-resolved brain responses to speech. We provide the Python code to replicate several validations analyses of the MEG evoked responses such as the temporal decoding of phonetic features and word frequency. All code and MEG, audio and text data are publicly available to keep with best practices in transparent and reproducible research.

Fig. 1

Dataset file structure.

Fig. 2

Median (across subjects) evoked response to all words. The gray area indicates the global field power (GFP).

Fig. 3

(a) Average (mean) decoding of whether the phoneme is voiced or not as a function of time following phoneme onset. The four colors refer to the four tasks (stories + word lists). Error bar are SEM across subjects. (b) Same as A for the decoding of words’ zipf frequency as a function of word onset. (c) Decoding of voicing (average across all tasks) for each participant, as a function of time following phoneme onset. (d) Same as C for decoding of word frequency (average across all tasks) for each participant, as a function of time following word onset.

Fig. 4

MEG data annotations: Pandas DataFrame of sound, phoneme and word time-stamps.