Neural encoding of acoustic features during speech and music perception: traduction of matlab code into python

Title

Neural encoding of acoustic and semantic features during speech and music perception: Matlab to Python code translation

Leaders

Bruno Giordano (INT) : https://github.com/brungio/bhack_td https://twitter.com/brungio https://framateam.org/blgnatsou/messages/@bruno.giordano

Giorgio Marinato (INT) : https://github.com/neurogima https://twitter.com/neurogima mattermost: @neurogima

Collaborators

Benjamin Morillon (CR INS - https://github.com/DCP-INS) Nadège Marin (IE, INS) Arnaud Zalta (PhD student, INS)

Brainhack Global 2022 Event

Brainhack Marseille

Project Description

In everyday life, humans are particularly attuned to listening to two particular types of sound: speech and music. We apply a novel analysis method to shed light on how the brain is almost effortlessly able to use acoustic features to assign meaning to sounds. To do so, we use an original cross-validated Representational Similarity Analyses (RSA) approach implemented in Matlab to estimate the similarity between acoustic or semantic features of an auditory stream (speech, music) and neural activity (here intracranial EEG recordings decomposed into frequency bands).

Link to project repository/sources

https://github.com/brungio/bhack_td

Goals for Brainhack Global

The main goal of this project is to translate the Matlab code into Python:

• translate the temporal folding on the neural signal in Python
• identify the python libraries to calculate different distances metrics
• translate the cross-validation method
• implement native plotting using the Python libraries (e.g. Seaborn, matplotlib, etc)

Good first issues

• identify python libraries that can speed up the code translation effort

Communication channels

https://mattermost.brainhack.org/brainhack/channels/brainhack_marseille_2022_speech_music_representation

Skills

python coding Representational Similarity Analyses (RSA) Cross-validation (train-test-validate) intracranial EEG signal processing sharing data analyses ideas

Onboarding documentation

What will participants learn?

to analyse intracranial EEG data to perform cross-validation procedures to estimate the neural encoding of different acoustic features

Data to use

We will provide a sample of a dataset of iEEG recordings.

Number of collaborators

from 3 to 10

Credit to collaborators

Acknowledgment in the code and in the Github repo.

Image

Leave this text if you don’t have an image yet.

Type

pipeline_development

Development status

1_basic structure

Topic

neural_decoding

Tools

MNE, RSAtoolbox (github.com/rsagroup/rsatoolbox), Scikit Learn

Programming language

Python, Matlab

Modalities

intracranial EEG

Git skills

Basic git workflow: fork, branches, commit and pull request