Speech Artifact Removal from Eeg Recordings of Spoken Word Production with Tensor Decomposition

Abstract

Research about brain activities involving spoken word production is considerably underdeveloped because of the undiscovered characteristics of speech artifacts, which contaminate electroencephalogram (EEG) signals and prevent the inspection of the underlying cognitive processes. To fuel further EEG research with speech production, a method using threemode tensor decomposition (time x space x frequency) is proposed to perform speech artifact removal. Tensor decomposition enables simultaneous inspection of multiple modes, which suits the multi-way nature of EEG data. In a picturenaming task, we collected raw data with speech artifacts by placing two electrodes near the mouth to record lip EMG. Based on our evaluation, which calculated the correlation values between grand-averaged speech artifacts and the lip EMG, tensor decomposition outperformed the former methods that were based on independent component analysis (ICA) and blind source separation (BSS), both in detecting speech artifact (0.985) and producing clean data (0.101). Our proposed method correctly preserved the components unrelated to speech, which was validated by computing the correlation value between the grand-averaged raw data without EOG and cleaned data before the speech onset (0.92-0.94).

Tongue Contour Tracking and Segmentation in Lingual Ultrasound for Speech Recognition: A Review

“…Although EEG can provide information about the speech patterns, the nature of the EEG signal is complex and susceptible to noise, which makes the part of the EEG complex signal relating to the auditory system difficult to be separated from other electrical activities of the brain [ 8 , 60 ]. Many advanced techniques have been proposed to alleviate this issue by proposing artifact removal [ 61 , 62 ] or incorporating advanced deep learning techniques such as a Transformer model and a generative adversarial network analysis [ 63 , 64 ].…”