SYSTEM AND METHOD FOR THE DISTRIBUTED SPEECH RECOGNITION USING THE SPEECH FEATURES

摘要

A network-based distributive voice recognition terminal, server, and system using voice features, and a method therefor are provided to carry out voice recognition by restoring compressed voice feature parameters received from the terminal, thereby reducing size(amount) of data necessary for voice recognition between the terminal and a voice recognition server. A voice input unit(11) inputs a voice spoken from a user, and a noise remover(12) removes an external noise signal mixed with user voice data. A voice feature generator(13) generates(extracts) voice feature parameters from the user voice data from which the noise signal is removed. A voice feature compressor(14) performs quantization for the voice feature parameters, and includes codebook indexes in the user voice data to compress the voice feature parameters. A voice feature transmitter(15) encodes the compressed voice feature parameters to send the encoded parameters to an external voice recognition server(20).