Abstract:Objective. To evaluate the potential of intracortical electrode array signals for brain-computer interfaces (BCIs) to restore lost speech, we measured the performance of decoders trained to discriminate a comprehensive basis set of 39 English phonemes and to synthesize speech sounds via a neural pattern matching method. We decoded neural correlates of spoken-out-loud words in the 'hand knob' area of precentral gyrus, a step toward the eventual goal of decoding attempted speech from ventral speech areas in patients who are unable to speak. Approach. Neural and audio data were recorded while two BrainGate2 pilot clinical trial participants, each with two chronically-implanted 96-electrode arrays, spoke 420 different words that broadly sampled English phonemes. Phoneme onsets were identified from audio recordings, and their identities were then classified from neural features consisting of each electrode's binned action potential counts or high-frequency local field potential power. Speech synthesis was performed using the 'Brain-to-Speech' pattern matching method. We also examined two potential confounds specific to decoding overt speech: acoustic contamination of neural signals and systematic differences in labeling different phonemes' onset times. Main results. A linear decoder achieved up to 29.3% classification accuracy (chance = 6%) across 39 phonemes, while an RNN classifier achieved 33.9% accuracy. Parameter sweeps indicated that performance did not saturate when adding more electrodes or more training data, and that accuracy improved when utilizing time-varying structure in the data. Microphonic contamination and phoneme onset differences modestly increased decoding accuracy, but could be mitigated by acoustic artifact subtraction and using a neural speech onset marker, respectively. Speech synthesis achieved r = 0.523 correlation between true and reconstructed audio. Significance. The ability to decode speech using intracortical electrode array signals from a nontraditional speech area suggests that placing electrode arrays in ventral speech areas is a promising direction for speech BCIs.

Individual Word Classification During Imagined Speech Using Intracranial Recordings

Imagined speech can be decoded from low- and cross-frequency intracranial EEG features

Towards Imagined Speech: Identification of Brain States from EEG Signals for BCI-based Communication Systems

Imagined speech event detection from electrocorticography and its transfer between speech modes and subjects

Decoding Single and Paired Phonemes Using 7T Functional MRI

Decoding spoken English phonemes from intracortical electrode arrays in dorsal precentral gyrus

Decoding Activity in Broca's Area Predicts the Occurrence of Auditory Hallucinations Across Subjects

Speech Activity Detection from Stereotactic EEG

Online internal speech decoding from single neurons in a human participant

Decoding speech perception from single cell activity in humans

Considerations for implanting speech brain computer interfaces based on functional magnetic resonance imaging

Performance of OpenBCI EEG Binary Intent Classification with Laryngeal Imagery

Decoding spoken English from intracortical electrode arrays in dorsal precentral gyrus

Imagined speech classification exploiting EEG power spectrum features

Speech imagery decoding as a window to speech planning and production

Intracranial Mapping of Response Latencies and Task Effects for Spoken Syllable Processing in the Human Brain

Imagined Speech and Visual Imagery as Intuitive Paradigms for Brain-Computer Interfaces

Continuous and discrete decoding of overt speech with electroencephalography

Resting state EEG assisted imagined vowel phonemes recognition by native and non-native speakers using brain connectivity measures

Online classification of imagined speech using functional near-infrared spectroscopy signals

Inner Speech Classification using EEG Signals: A Deep Learning Approach