Hevolve AI: Self-Evolving Multimodal AI Agents

Turn your domain expertise into AI agents that keep learning. Hevolve AI lets experts build multimodal AI systems by talking to them and correcting them in real time, with no code to write.

Key Features

Quick Links

© 2024 Hevolve AI Pvt Ltd. All rights reserved.

← All research
Brain–Computer Interfaces
arXiv (BCI) · July 4, 2026

EEG-Based Imagined Speech Decoding Using a Hybrid CNN-SNN Architecture

Fatima Shalhoub, Mariam Al Mawla, Kabalan Chaccour +2 more

Imagined speech decoding tries to read the words you say silently in your head from EEG, which would be life-changing for people who can't speak. It's brutally hard, though: the signals are faint, jittery, and never quite the same twice. Most decoders lean on standard machine learning that ignores the fine timing of when neurons fire, which is exactly where spiking neural networks feel at home, since they process events in time the way biological neurons do.

This paper stitches the two together. A convolutional network first pulls out temporal features, then a spiking network does the timing-aware classification. On a 2020 BCI Competition benchmark it reached about 80% accuracy, beating prior methods that topped out near 70%, and the authors say it's the first to bring SNNs to imagined-speech decoding.

That's from the abstract, so read the paper for how many classes and subjects were involved before weighing the 80%.

From the arXiv (BCI) abstract

Imagined speech decoding using EEG signals has emerged as a promising frontier in brain-computer interface (BCI) research, particularly to restore communication for individuals with severe speech impairments. However, decoding imagined speech remains a complex task due to the non-stationary, low-amplitude, and highly variable nature of EEG signals. Existing methods often rely on classical machine learning or deep learning models that fail to exploit spike-based temporal…


More Brain–Computer Interfaces papers