Text Independent Speaker Verification Using Adapted Gaussian Mixture Models Textoberoende talarverifiering med adapterade Gaussian-Mixture-modeller

نویسنده

  • Daniel Neiberg
چکیده

The primary goal of this master thesis project is to implement a text independent speaker verification module for GIVES. Secondary goals are to implement a fast scoring method and compare performance between the implemented text independent module and an available text dependent module. The project also includes a literature study. The text independent module is based on adapted Gaussian Mixture Models and the adaptation equations are derived. Evaluation results show that the text independent module and the text dependent module have almost equal performance on a text dependent recognition task. The results are analyzed and summarized, and improvements are suggested. Unfortunately, the fast scoring method did not work together with all the components in GIVES. Sammanfattning Det primära målet med detta examensarbete är att implementera en textoberoende talarverifieringsmodul för GIVES. Sekundära mål är att implementera en snabb verifieringsmetod och att jämföra prestanda mellan den implementerade textoberoende modulen och en befintlig textberoende modul. Examensarbetet inkluderar ocks̊a en litteraturstudie. Den textoberoende modulen baseras p̊a adapterade Gaussian-Mixture-modeller och adapteringsekvationerna härleds. En utvärdering visar att den textoberoende modulen och den textberoende modulen har likvärdiga prestanda p̊a en textberoende igenkänningsuppgift. Resultaten analyseras och summeras, och förbättringar föresl̊as. Tyvärr s̊a fungerade inte den snabba verifieringsmetoden med alla komponenterna i GIVES.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Text-Independent Speaker Recognition Using Gaussian Mixture Models Final Term Paper Proposal

The proposed project is an implementation of speaker recognition systems, both identification and verification. The systems are built using Gaussian Mixture Models, as proposed in several papers from Douglas A. Reynolds. The use of Fractional Covariance Matrix is studied as an possible increase for the traditional recognition systems. keywords: speaker recognition; Gaussian Mixture Models; like...

متن کامل

Speaker Verification System Using Gaussian Mixture Model & UBM

In This paper presents an overview of a stateof-the-art text-independent speaker verification system. The objective of automatic speaker recognition is to extract, characterize and recognize the information about speaker identity. First, an introduction proposes a modular scheme of the training and test phases of a speaker verification system. Then, the most commonly speech parameterization use...

متن کامل

Speaker Verification Using Adapted Gaussian Mixture Models

In this paper we describe the major elements of MIT Lincoln Laboratory’s Gaussian mixture model (GMM)-based speaker verification system used successfully in several NIST Speaker Recognition Evaluations (SREs). The system is built around the likelihood ratio test for verification, using simple but effective GMMs for likelihood functions, a universal background model (UBM) for alternative speaker...

متن کامل

New background modeling for speaker verification

A new background speaker modelling method is presented in this paper for text-independent speaker verification using Gaussian mixture models. This method does not require speech databases of other speakers to build background speaker models. A background model can be built directly from the same claimed speaker's database and has a smaller number of Gaussian mixtures compared to the claimed spe...

متن کامل

Discriminative Transformation for Sufficient Adaptation in Text-Independent Speaker Verification

In conventional Gaussian Mixture Model – Universal Background Model (GMM-UBM) text-independent speaker verification applications, the discriminability between speaker models and the universal background model (UBM) is crucial to system’s performance. In this paper, we present a method based on heteroscedastic linear discriminant analysis (HLDA) that can enhance the discriminability between spea...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2001