Related Experiment Video
Updated: May 20, 2026
05:48
Memorization-Based Training and Testing Paradigm for Robust Vocal Identity Recognition in Expressive Speech Using Event-Related Potentials Analysis
Published on: August 9, 2024
EffortNet: A Deep Learning Framework for Objective Assessment of Speech Enhancement Technologies Using EEG-Based
Summary
EffortNet, a new deep learning model, decodes listening effort from electroencephalography (EEG) signals. It accurately measures listening effort in individuals, outperforming other models for hearing technology evaluation.
Area of Science:
- Neuroscience
- Speech and Hearing Science
- Artificial Intelligence
Background:
- Quantifying listening effort during speech comprehension is crucial but challenging.
- Individual variability in electroencephalography (EEG) signals complicates effort decoding.
- Alpha power in EEG is a validated objective biomarker for listening effort.
Purpose of the Study:
- To introduce EffortNet, a deep learning framework for individual-level listening effort decoding using EEG.
- To evaluate EffortNet's performance against conventional models.
- To assess the neural impact of different speech enhancement technologies.
Main Methods:
- Collected 64-channel EEG data from 122 participants during speech comprehension tasks.
- Utilized a deep learning framework integrating self-supervised, incremental, and transfer learning.
- Compared EffortNet with Convolutional Neural Networks (CNN) and STAnet models.
Main Results:
- EffortNet achieved 80.9% accuracy in decoding listening effort, significantly outperforming CNN (62.3%) and STAnet (61.1%).
- The model demonstrated high performance with limited training data from new subjects (40%).
- Transformer-enhanced speech showed neural responses closer to clean speech than MMSE-enhanced speech.
Conclusions:
- EffortNet provides an accurate and individualized method for quantifying listening effort.
- Objective neural markers from EffortNet offer valuable insights into hearing technology performance.
- Dissociation between neural responses and subjective ratings highlights the importance of objective measures.
