An integrated system for text-independent speaker recognition using binary neural network classifiers

Abstract

Speaker recognition consists of speaker identification and speaker verification. Many speaker recognition systems can only perform either the identification or the verification task. This paper is intended to investigate an integrated text-independent speaker recognition system, which is suitable for both identification and verification. One obvious advantage of such an integrated system is that it simplifies the big classification problem because it combines a series of binary neural network classifiers (BNNCs), each of which classifies only two speakers. A novel usage of the cohort normalization method is presented in this system which makes it easier to perform the verification task. Experiments show that this system performs well for both identification and verification tasks.

Related Papers

• → Hierarchical speaker identification using speaker clustering(2004)24 cited
• → A speaker identification system with verification method based on speaker relative threshold and HMM(2003)4 cited
• → Speaker Recognition and Diarization(2010)3 cited
• A Semi-Automatic Approach to Create Large Gender-and Age-Balanced Speaker Corpora: Usefulness of Speaker Diarization & Identification(2022)
• → Fixed-Phrase Speaker Verification(2011)