Speaker Recognition with Deep Learning Approaches: A Review

This article gives an overview of the methods for using deep learning to identify and verify speakers. Speaker recognition is an everyday use of speech technology. Many research initiatives have been carried out in the past few years, but little progress has been achieved. But just as deep learning techniques are replacing previous state-of-the-art approaches in speech recognition, they are also developing in most machine learning fields. Deep learning seems to have evolved into the most advanced speaker verification and identification technique. Most novel efforts start with the common x-vectors in addition to i-vectors. The increasing volume of data gathered makes the area where deep learning is most effective more accessible.