Brain-computer interface system and method for recognizing conversation intention of user using the same

A method for recognizing a conversational intention of a user by a brain-computer interface system that recognizes the intention of a user through a brainwave, the method comprising:

• a) measuring the brainwave for a conversation sentence intended by the user through a sensor module of detecting the brainwave of the user in real time,
• b) classifying the candidate words to be estimated as the word intended by the user from the measured brainwave based on a learned decoding model;
• and
• c) providing a result obtained by combining the classified candidate words,
• wherein the learned decoding model is generated to classify the user's intentions based on the features of the brain waves generated when performing imagined speech or audible act and sound waves generated when performing speech act, and
• wherein the learned decoding model includes a classification model of classifying the words intended by the user through features for the brainwave and the soundwave generated when performing the imagined speech, the speech act, or the audible act, and a soundwave provision model of deriving correlation by extracting envelope between the brainwave data and the soundwave data and then providing a soundwave waveform similar to that of the brainwave through the derived correlation.

A brain-computer interface system for recognizing a conversational intention of a user through a brainwave comprising:

• a sensor module that detects a brainwave of a user;
• a memory that stores a program for executing a method for recognizing the conversational intention of a user by using the brainwave detected in the sensor module;
• and
• a processor that executes the program,
• wherein by executing the program, the processor measures the brainwave for a conversation sentence intended by the user through the sensor module in real time, classifies the candidate words to be estimated as the word intended by the user from the measured brainwave based on a learned decoding model, and provides a result obtained by combining the classified candidate words, and
• wherein the learned decoding model is generated to classify the user's intentions based on the features of the brain waves generated when performing imagined speech or audible act and sound waves generated when performing speech act, and
• wherein the learned decoding model includes a classification model of classifying the words intended by the user through features for the brainwave and the soundwave generated when performing the imagined speech, the speech act, or the audible act, and a soundwave provision model of deriving correlation by extracting envelope between the brainwave data and the soundwave data and then providing a soundwave waveform similar to that of the brainwave through the derived correlation.