Brain Activity Chord Decoding via Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies lack the ability to directly decode chord information from brain activity, resulting in low accuracy and information loss when reconstructing musical stimuli and using automatic chord estimation methods, which hinders applications in healthcare and musical creation.
Innovation Solution
A deep learning-based system that uses functional neuroimaging to measure brain activity, extracts patterns, and employs a decoding model to directly convert brain activity into chord information, bypassing the need for musical stimuli reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If auditory stimuli decoding technology is used to reconstruct musical stimuli from brain activity, then musical stimuli can be recovered, but information loss occurs during reconstruction
Solution Approach 1:
The patent extracts only the necessary chord information directly from brain activity patterns using a decoding model, rather than attempting to reconstruct the complete musical stimulus. This selective extraction approach obtains chord labels (root note and chord quality) directly from neural patterns, bypassing the information loss inherent in full stimulus reconstruction.
Solution Approach 2:
Instead of following the conventional path of reconstructing musical stimuli first and then estimating chords, the patent inverts the process by directly decoding chord information from brain activity patterns. This reverse approach eliminates the intermediate reconstruction step that causes information loss.
2Measurement precision
If automatic chord estimation technology is used on reconstructed music, then chord information can be obtained, but secondary information loss occurs
Solution Approach 1:
The patent extracts chord information directly from brain activity patterns through a trained decoding model, eliminating the need for automatic chord estimation on reconstructed music. This direct extraction approach avoids the secondary information loss that occurs when chord estimation is applied to already-reconstructed musical stimuli.
Solution Approach 2:
The patent inverts the conventional chord estimation pipeline by decoding chord information directly from neural patterns rather than from reconstructed audio signals. This inversion removes the intermediate reconstruction-estimation steps that cause cumulative information loss.
3Measurement precision
If conventional ACE-based methods are used for inner music, then chord estimation can be performed, but accuracy is limited due to lack of audio signals
Solution Approach 1:
The patent replaces the mechanical/audio-based chord estimation system with a neuroimaging-based decoding system. Instead of requiring audio signals as input for ACE, the system uses brain activity patterns (fMRI signals) as input, substituting the sensory modality from auditory to neural imaging while maintaining chord estimation functionality.
Solution Approach 2:
The patent introduces brain activity patterns as an intermediary between inner music experiences and chord information extraction. Rather than attempting to access inner music directly or require external audio, the system uses neural patterns as a mediator that preserves chord information from subjective musical experiences.
Data Source
AI summary
Disclosed are systems and methods for decoding chord information from brain activity. General chord decoding protocols involves using computational operations for the extraction of neural codes, the development of the decoding model, and the deployment of the trained model.