Deep belief networks for i-vector based speaker recognition

Abstract

The use of Deep Belief Networks (DBNs) is proposed in this paper to model discriminatively target and impostor i-vectors in a speaker verification task. The authors propose to adapt the network parameters of each speaker from a background model, which will be referred to as Universal DBN (UDBN). It is also suggested to backpropagate class errors up to only one layer for few iterations before to train the network. Additionally, an impostor selection method is introduced which helps the DBN to outperform the cosine distance classifier. The evaluation is performed on the core test condition of the NIST SRE 2006 corpora, and it is shown that 10% and 8% relative improvements of EER and minDCF can be achieved, respectively.Peer ReviewedPostprint (published version

From Features to Speaker Vectors by means of Restricted Boltzmann Machine Adaptation

“…In other words, the URBM is adapted to the data of each speaker. The idea of this kind of adaptation has also shown success in [11,12,13,14] to initialize the parameters of DNNs for classification purposes. Figure 3 shows the weight matrices for URBM along with its adapted versions for two randomly selected speakers.…”

“…They have been utilized in an adaptation process [11,12,13,14], to further discriminatively model target and impostor speakers. RBMs have been recently used in DBNs as a pre-training stage to extract Baum-Welch statistics for i-vector and supervector extraction [15,16].…”

Deep Belief Networks for Fingerprinting Indoor Localization Using Ultrawideband Technology

“…As a typical deep learning model, DBN solves the training problem which may occur in a deep neural network. It is widely used in many different areas in the recent years, such as graphics processing and language recognition [12][13][14]. DBN is advanced model which can fit the complex nonlinear relationship between attributes in many issues [15,16].…”

“…It has been widely used in many different fields such as image [12], speech [13], and language processing [14]. It is a multilayer model which imitates the mode in which the human brain represents information.…”

Multiple Discriminative DNNs for I-Vector Based Open-Set Language Recognition

“…이 뿐만 아니라, 입력 시퀀스와 출력 라벨 사이의 비선형적 관계를 표현할 수 있는 DNN의 능력을 활용하기 위하여 DNN을 화자 인식에서 인식 기로 직접 사용하는 방법에 대한 연구도 진행되어왔 다. [4,5] 화자 인식에서 DNN을 이용한 분류 기법은 기 존에 사용되어온 support vector machine (SVM) 및 cosine distance 기반의 분류 방식에 비하여 높은 성능 을 보였다. 화자 인식에서와 마찬가지로 DNN은 연령 인식 [6,7] 과 언어 인식에서도 높은 성능을 보였다.…”