Music genre neuroimaging dataset

Area of Science:

• Auditory Neuroscience
• Neuroimaging
• Machine Learning
• Computational Neuroscience

Background:

• Functional magnetic resonance imaging (fMRI) data were collected from five subjects.
• Subjects listened to 540 music pieces across 10 genres over 3 days.
• Behavioral data are also available, alongside training and test data splits.

Purpose of the Study:

• To provide a dataset for testing machine learning algorithms on predicting brain activity from music stimuli.
• To facilitate research integrating auditory neuroscience and machine learning.
• To enable evaluation of signal-to-noise ratio in brain activity using repeated stimuli.

Main Methods:

• fMRI data acquisition during music listening.
• Collection of behavioral responses.
• Dataset partitioning into training and testing sets.
• Repetition of test stimuli for reliability assessment.

Main Results:

• The dataset enables testing of prediction performance of brain activity.
• Allows for evaluation of various machine learning algorithms.
• Facilitates the application of novel acoustic models.

Conclusions:

• This dataset supports the development and validation of computational models in auditory neuroscience.
• It bridges the gap between neuroimaging and machine learning research.
• Further research can explore advanced acoustic feature extraction and predictive modeling.