METHOD AND SYSTEM FOR RECOGNIZING SPEAKER UTILIZING AGE INFORMATION

摘要

According to an embodiment of the present invention, a method for recognizing a speaker capable of reducing misidentification between family members comprises the steps of: inputting a plurality of registration voices corresponding to a plurality of speakers; generating a plurality of speaker models corresponding to the speakers by applying an age estimation model to each of the registration voices; estimating age of each of the speakers using the age estimation model; registering the speaker models and estimated ages; inputting a novel voice; setting a candidate group consisting of part of the speakers using the novel voice and estimated ages; and identifying a speaker corresponding to the novel voice from the candidate group.;COPYRIGHT KIPO 2020