BCIwiki (bciwiki.com) — Speech brain-computer interfaces (BCIs) should be treated as adaptive, user-centered clinical systems rather than neural-to-text decoders, and judged on more than offline accuracy, according to a review posted to arXiv on September 29, 2026. The paper is a preprint and has not been peer reviewed. Its authors are Moein Khajehnejad, Forough Habibollahi, Tommaso Boccato, Margarida Sousa, Michal Olak, Francesco Jamal Sheiban and Matteo Ferrante; the review runs 28 pages with 4 figures, 2 boxes and 2 tables.
The authors describe speech BCIs as aiming to restore communication by transforming neural activity related to speech, language or communicative intent into outputs such as text, synthesized voice or avatar control. They credit advances in intracortical and electrocorticographic recording, deep sequence models and language-model-assisted decoding with enabling high-performance attempted-speech decoding and increasingly naturalistic speech synthesis. In practice, they write, neural representations, recording hardware, decoding architectures, language priors, feedback and user learning interact over time.
The review first examines the neural substrates of speech and language, emphasizing their hierarchical, distributed, temporally structured and non-stationary organization, then covers recording and decoding choices, closed-loop adaptation, evaluation, clinical translation and ethics. It highlights recurring trade-offs between signal resolution and invasiveness, low-level motor and high-level semantic targets, decoder accuracy and user agency, and language-model fluency and faithful neural evidence.
The authors argue the next generation of speech BCIs should also be assessed on robustness across sessions, calibration burden, latency, the ability to express uncertainty, usability and safeguards against unintended decoding. They outline priorities spanning speech neuroscience, neural engineering, machine learning, clinical practice and neuroethics needed to move from proof-of-concept decoding toward reliable, expressive and controllable communication neuroprostheses.