# Advancing Speech Recognition With No Speech Or With Noisy Speech

## TL;DR

This paper explores novel speech recognition methods that utilize EEG signals without speech input and improve recognition in noisy environments by combining EEG and speech data.

## Contribution

It introduces end-to-end CSR models using EEG signals alone and demonstrates fusion of EEG with noisy speech for enhanced recognition accuracy.

## Key findings

- EEG-based CSR achieves recognition without speech signals.
- Fusion of EEG and noisy speech improves accuracy.
- End-to-end models outperform traditional methods.

## Abstract

In this paper we demonstrate end-to-end continuous speech recognition (CSR) using electroencephalography (EEG) signals with no speech signal as input. An attention model based automatic speech recognition (ASR) and connectionist temporal classification (CTC) based ASR systems were implemented for performing recognition. We further demonstrate CSR for noisy speech by fusing with EEG features.