Analysis of simultaneous recording of EEG and speech signals



Journal Title

Journal ISSN

Volume Title



Speech recognition or Automatic Speech Recognition System ( ASR) is a system which converts speech into text. ASR systems forms back end in many cutting edge technologies used by Apple, Google, Amazon, etc in today’s world. Some examples of systems which uses ASR in front end or back end include Apple’s IOS Siri, Amazon’s Alexa, Google’s talk, Microsoft’s Cortana etc. The main problem with this type of systems is that they don’t perform well in presence of background noise. This lead to the design of robust ASR systems. Speech recognition is a very challenging problem as one has to deal with background noise, variation with rate of speech, variation with age of the speaker etc. All this factors affects the accuracy of the ASR systems. Recently the use of EEG or electroencephalography in Speech recognition is gaining popularity among research community. EEG has been used to perform attention detection in cocktail party problem to identify the right speaker. In this report the author focuses on the study of the effect of EEG on vocalized and silent speech. A detailed analysis of speech and EEG signals is provided in this report. The main objective of the work explained in this report was to identify the location and number of EEG sensors required to accurately map the activity of neurons firing during speech activation and how the information from these neurons may be used to improve the accuracy of an ASR system.


LCSH Subject Headings