This paper describes text-independent speaker recognition system which is basic and essential for interaction between users and the robot in Ubiquitous Robot Companion environment. For comfortable interaction between users and the robot, we implement an online speaker enrollment module which trains a GMM for each speaker model and text-independent speaker recognition module which consists of te...

The goal of the SpeechDat project is to develop spoken language resources for speech recognisers suited to realise voice driven teleservices. SpeechDat created speech databases for all official languages of the European Union and some major dialectal varieties and minority languages. The size of the databases ranges between 500 and 5000 speakers. In total 20 databases are recorded over the fixe...

One approach to ASR is based on the Hearsay II (ref 1) architecture which utilises a nurober of phonetic and linguistic knowledge sources in its recognition process. The fundamental ideas behind the Hearsay II architecture has been used with a Danish speech project, which is aiming for industrial applications implemented into a Standard microprocessor system to demonstrate a continuous speech, ...

This paper is about different methods and algorithms that were used for speaker identification from the video recordings of TV broadcast news transcription. The information from visual speaker identification were used in our complex system for automatic continuous speech recognition of TV broadcast programs because it is possible to use speaker adapted (SA) Hidden Markov Models (HMMs) if we hav...

This paper aims at speaker recognition based upon a novel set of features. Feature extraction is a crucial phase of the speaker recognition process and a proper feature set can influence it dramatically. Many well-known features are not suitable for the speaker recognition as those merge the specifics of the individual voices to make them universal. Therefore, we need features accentuating the ...

Automatic speaker recognition on a vocoder link has rarely been explicitly tested. In this paper, we show how the automatic speaker recognition could be used on a vocoder link. In a first experiment where we consider the “coder-link-decoder” speech system as a black box, a classic speaker recognition method (applied on the reconstructed speech) is shown to be able to provide an objective measur...

This paper aims at providing a brief overview into the area of speaker recognition. Speaker recognition can be classified into text dependent and the text independent methods. The features of speech signal that are being used (or have been used) for speaker recognition are presented briefly in this paper. The commonly used techniques of pattern matching for the decision making process in speake...

In this Thesis two wavelet based methods (TADWT and SSWPT) for enhancement of noisy speech are developed and reported. The effectiveness of the reported speech enhancement techniques is evaluated by means of speaker recognition under noisy conditions. The automatic speaker recognition technologies have been developed into more and more important modern technologies required by many speech-aided...

Recently, Deep Neural Network (DNN) based bottleneck features proved to be very effective in i-vector based speaker recognition. However, the bottleneck feature extraction is usually fully optimized for speech rather than speaker recognition task. In this paper, we explore whether DNNs suboptimal for speech recognition can provide better bottleneck features for speaker recognition. We experimen...

The newly collected Speakers in the Wild (SITW) database was central to a text-independent speaker recognition challenge held as part of a special session at Interspeech 2016. The SITW database is composed of audio recordings from 299 speakers collected from open source media, with an average of 8 sessions per speaker. The recordings contain unconstrained or “wild” acoustic conditions, rarely f...