Filters








706 Hits in 5.9 sec

Speaker model adaptation based on confidence score

Erhan Mengusoglu
2015 Tehnički Vjesnik  
A linear interpretation of confidence measure is very important to select the most representative data for adaptation.  ...  In this paper we introduced a linear interpretation of posterior probability based confidence measure by using inverse Fisher transformation.  ...  Maximum Likelihood Linear Regression (MLLR) The use of MLLR for model adaptation consists in producing a set of regression based transforms from some adaptation data.  ... 
doi:10.17559/tv-20140120095957 fatcat:ns44x3pyj5cq5bp7pdigjea3f4

Domain-invariant I-vector Feature Extraction for PLDA Speaker Verification

Md Hafizur Rahman, Ivan Himawan, David Dean, Clinton Fookes, Sridha Sridharan
2018 Odyssey 2018 The Speaker and Language Recognition Workshop  
The performance of the current state-of-the-art i-vector based probabilistic linear discriminant analysis (PLDA) speaker verification depends on large volumes of training data, ideally in the target domain  ...  In this paper, we introduce a domain-invariant i-vector extraction (DI-IVEC) approach to extract domain mismatch compensated out-domain i-vectors using limited in-domain (target) data for adaptation.  ...  They used maximum likelihood linear transformation (MLLT) technique to estimate the transfer parameters and expectation maximization (EM) to get the adapted PLDA parameters.  ... 
doi:10.21437/odyssey.2018-22 dblp:conf/odyssey/RahmanHDFS18 fatcat:sl3o3lkasfa47jyknijtoqvjya

Spoofing-Aware Speaker Verification with Unsupervised Domain Adaptation [article]

Xuechen Liu, Md Sahidullah, Tomi Kinnunen
2022 arXiv   pre-print
We employ three unsupervised domain adaptation techniques to optimize the back-end using the audio data in the training partition of the ASVspoof 2019 dataset.  ...  In this paper, we initiate the concern of enhancing the spoofing robustness of the automatic speaker verification (ASV) system, without the primary presence of a separate countermeasure module.  ...  Unsupervised Domain Adaptation One of the key challenges in speaker recognition is generalization across different domains.  ... 
arXiv:2203.10992v2 fatcat:dn56nhmshnhzbkzdev7x3diaam

A Generalized Framework for Domain Adaptation of PLDA in Speaker Recognition [article]

Qiongqiong Wang, Koji Okabe, Kong Aik Lee, Takafumi Koshinaka
2020 arXiv   pre-print
This paper proposes a generalized framework for domain adaptation of Probabilistic Linear Discriminant Analysis (PLDA) in speaker recognition.  ...  It not only includes several existing supervised and unsupervised domain adaptation methods but also makes possible more flexible usage of available data in different domains.  ...  A linear interpolation method has been proposed for combining parameters of PLDAs trained separately with OOD and InD data so as to take advantage of both PLDAs [13] ; in [20] , a maximum likelihood  ... 
arXiv:2008.08815v1 fatcat:webto7nqk5gqpdhyq7gdwowu7e

Weakly Supervised PLDA Training [article]

Lantian Li, Yixiang Chen, Dong Wang, Chenghui Zhao
2017 arXiv   pre-print
PLDA is a popular normalization approach for the i-vector model, and it has delivered state-of-the-art performance in speaker verification.  ...  We present a cheap PLDA training approach, which assumes that speakers in the same session can be easily separated, and speakers in different sessions are simply different.  ...  [7] proposed a domain-adaptation approach based on maximum likelihood linear transformation (MLLT), and Rahman et al.  ... 
arXiv:1609.08441v2 fatcat:odryucpaajavlct4grirlle3iu

Environment adaptation for robust speaker verification by cascading maximum likelihood linear regression and reinforced learning

K.K. Yiu, M.W. Mak, S.Y. Kung
2007 Computer Speech and Language  
are used for verification.  ...  In speaker verification over public telephone networks, utterances can be obtained from different types of handsets.  ...  Maximum Likelihood Linear Regression MLLR was originally developed for speaker adaptation [6] ; however, it can also be applied to environment adaptation.  ... 
doi:10.1016/j.csl.2006.05.001 fatcat:og7qtepnpfgxzih53o4qje4zua

Adaptation Algorithms for Speech Recognition: An Overview [article]

Peter Bell, Joachim Fainberg, Ondrej Klejch, Jinyu Li, Steve Renals, Pawel Swietojanski
2020 arXiv   pre-print
systems, with a focus on speaker adaptation, domain adaptation, and accent adaptation.  ...  We present a meta-analysis of the performance of speech recognition adaptation algorithms, based on relative error rate reductions as reported in the literature.  ...  In the mid-1990s, the influential maximum likelihood linear regression (MLLR) [78] and maximum a posteriori (MAP) [79] approaches to speaker adaptation for HMM/GMM systems were introduced.  ... 
arXiv:2008.06580v1 fatcat:7cukuwdfjvdtpdnxb6gdmfywbu

Local Training for PLDA in Speaker Verification [article]

Chenghui Zhao, Lantian Li, Dong Wang, April Pu
2016 arXiv   pre-print
PLDA is a popular normalization approach for the i-vector model, and it has delivered state-of-the-art performance in speaker verification.  ...  A possible approach to mitigate the problem is various unsupervised adaptation methods, which use unlabeled data to adapt the PLDA scattering matrices to the target domain.  ...  [11] proposed a domain-adaptation approach based on maximum likelihood linear transformation (MLLT), and Rahman et al.  ... 
arXiv:1609.08433v1 fatcat:yq7zl7dxh5ae3difb6spknuvia

Adaptation Algorithms for Speech Recognition: An Overview

Peter Bell, Joachim Fainberg, Ondrej Klejch, Jinyu Li, Steve Renals, Pawel Swietojanski
2020 IEEE Open Journal of Signal Processing  
systems, with a focus on speaker adaptation, domain adaptation, and accent adaptation.  ...  We present a meta-analysis of the performance of speech recognition adaptation algorithms, based on relative error rate reductions as reported in the literature.  ...  In the mid-1990 s, the influential maximum likelihood linear regression (MLLR) [82] and maximum a posteriori (MAP) [83] approaches to speaker adaptation for HMM/GMM systems were introduced.  ... 
doi:10.1109/ojsp.2020.3045349 fatcat:tdprznnisvc2xampkzbavjpamu

Table of Contents

2018 IEEE/ACM Transactions on Audio Speech and Language Processing  
Fazi 1539 Maximum-Likelihood Linear Transformation for Unsupervised Domain Adaptation in Speaker Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ...  Domain Adaptive Blind Multichannel Identification With p -Norm Constraint . . . . . . . . . . . . . .  ... 
doi:10.1109/taslp.2018.2855923 fatcat:suvdeywybva35l577w2lvenizi

VAE-based Domain Adaptation for Speaker Verification [article]

Xueyi Wang and Lantian Li and Dong Wang
2019 arXiv   pre-print
Deep speaker embedding has achieved satisfactory performance in speaker verification.  ...  Our experiments demonstrated that by this VAE-adaptation approach, speaker embeddings can be easily transformed to the target domain, leading to noticeable performance improvement.  ...  It factorizes the speech signal into the phonetic factor and the speaker factor, and this factorization process is based on the maximum likelihood (ML) criterion.  ... 
arXiv:1908.10092v1 fatcat:mosxtdhfxff5xoogcru3ravogy

Unsupervised training of an HMM-based self-organizing unit recognizer with applications to topic classification and keyword discovery

Man-hung Siu, Herbert Gish, Arthur Chan, William Belfield, Steve Lowe
2014 Computer Speech and Language  
Abstract We present our approach to unsupervised training of speech recognizers. Our approach iteratively adjusts sound units that are optimized for the acoustic domain of interest.  ...  We thus enable the use of speech recognizers for applications in speech domains where transcriptions do not exist. The resulting recognizer is a state-of-the-art recognizer on the optimized units.  ...  Acknowledgments We would like to thank the anonymous reviewers for their useful comments and suggestions.  ... 
doi:10.1016/j.csl.2013.05.002 fatcat:7bwh437smbgcphmb42fosuicoe

The Likelihood Ratio Decision Criterion for Nuisance Attribute Projection in GMM Speaker Verification

Boštjan Vesnicer, France Mihelič
2008 EURASIP Journal on Advances in Signal Processing  
We propose a way of integrating likelihood ratio (LR) decision criterion with nuisance attribute projection (NAP) for Gaussian mixture model-(GMM-) based speaker verification.  ...  The experiments on the core test of the NIST speaker recognition evaluation (SRE) 2005 data show that the performance of the proposed approach is comparable to that of the standard approach of NAP which  ...  The LR-based decision criterion If the decision criterion is based on the likelihood ratio, then the verification score is calculated as (In practice, the likelihood ratio is computed in the log domain  ... 
doi:10.1155/2008/786431 fatcat:svxfbo7u4vagbpsyt6cbucz56y

UIAI System for Short-Duration Speaker Verification Challenge 2020 [article]

Md Sahidullah, Achintya Kumar Sarkar, Ville Vestman, Xuechen Liu, Romain Serizel, Tomi Kinnunen, Zheng-Hua Tan, Emmanuel Vincent
2020 arXiv   pre-print
In this work, we present the system description of the UIAI entry for the short-duration speaker verification (SdSV) challenge 2020.  ...  We investigate different feature extraction and modeling approaches for automatic speaker verification (ASV) and utterance verification (UV).  ...  The best-matched PBM is found in the maximum likelihood (ML) sense. In the test phase, we first determine the best-matched PBM for the test utterance.  ... 
arXiv:2007.13118v1 fatcat:vs47wxsar5gk3eojsrpgizdhie

Estimation of Noise Power Spectrum and Automatic Speaker Recognition System

Rajalakshmi P, Anju L
2017 IJARCCE  
Then the enhanced speech signal is obtained by transforming the estimated spectrum into time domain.  ...  The results of unsupervised samples are similar to the supervised one. This method is highly efficient for learning real world datasets.  ...  GMM is adapted from a Universal Background Model (UBM) and is currently the most popular approach for speaker verification.  ... 
doi:10.17148/ijarcce.2017.6575 fatcat:3n63i5geqfearocbwrn4fmtigm
« Previous Showing results 1 — 15 out of 706 results