A Generalised Normalisation Method for Speaker Veri ̄cation
نویسندگان
چکیده
In a speaker veri ̄cation system, a claimed speaker's score is computed to accept or reject the speaker claim. Most of the current methods compute the score as the ratio of the claimed speaker's and the impostors' likelihood functions. Based on analysing false acceptance and false rejection errors obtained by using these methods, we propose a generalised method to ̄nd better scores which can reduce both error types. Proposed scores are the ratios of the functions of the claimed speaker's and the impostors' likelihood functions. Experiments performed on the ANDOSL and YOHO speech corpora show better results for the proposed method.
منابع مشابه
Fuzzy C-Means Clustering-Based Speaker Veri cation
In speaker veri cation, a claimed speaker's score is computed to accept or reject the speaker claim. Most of the current normalisation methods compute the score as the ratio of the claimed speaker's and the impostors' likelihood functions. Based on analysing false acceptance error occured by the current methods, we propose a fuzzy c-means clusteringbased normalisation method to nd a better scor...
متن کاملFuzzy normalisation methods for speaker verification
This paper proposes normalisation methods based on fuzzy set theory for speaker veri cation. A claimed speaker's score used to accept or reject this speaker is viewed as a fuzzy membership function. We propose two scores: the fuzzy entropy and fuzzy C-means membership functions. Moreover, a likelihood transformation is considered to obtain a general approach and, based on this, ve more fuzzy sc...
متن کاملAdvances on HMM-based text-dependent speaker verification
This paper presents recent development on text-dependent speaker verification technology in EU project PICASSO, which have improved the SV performance significantly. In the project we adopt HMM approach for pattern matching. In the paper we describes four different techniques, adaptive variance flooring, multiple use of enrolment sample, generalised competitive measurement for score normalisati...
متن کاملText-independent speaker recognition using non-linear frame likelihood transformation
When the reference speakers are represented by Gaussian mixture model (GMM), the conventional approach is to accumulate the frame likelihoods over the whole test utterance and compare the results as in speaker identi®cation or apply a threshold as in speaker veri®cation. In this paper we describe a method, where frame likelihoods are transformed into new scores according to some non-linear func...
متن کاملField test of a calling card service based on speaker verification and automatic speech recognition
In this research we have studied several human factors problems that are connected to the deployment of speaker veri cation technology in telecommunication services. We investigate the perception of the safety of a calling card service when it is protected by speaker veri cation on the 14 digit card number, and compare it to the perceived safety of speaker veri cation and PIN. Moreover, we comp...
متن کامل