ICITR - 2019

Permanent URI for this collectionhttp://192.248.9.226/handle/123/14731

Browse

Browsing ICITR - 2019 by Author "Amalraj, CRJ"

Filter results by typing the first few letters
Now showing 1 - 3 of 3
  • No Thumbnail Available
    item: Conference-Full-text
    Feature based speaker embedding on conversational speeches
    (Information Technology Research Unit, Faculty of Information Technology, University of Moratuwa, Sri Lanka, 2019-12) Balasubramaniyam, H; Amalraj, CRJ; Sudantha, BH
    Identifying the speaker of a specific speech by examining the speech features of the voice is called speaker identification. The task of speaker identification consists of three main phases which are feature extraction, feature embedding and voice classification. Speaker embedding is the process of modeling the voice of a person where the model of the utterance can uniquely represent the speaker of that voice. Speaker embedding is a commonly used method in Automatic Speaker Recognition systems to identify the voice of the speaker. Currently, Deep Neural Networks based approaches are used in these systems for speech feature extraction and speech embedding. The performances of different approaches heavily depend on the noise factor and suitability of selected features of the audio data. MFCC, LPC, Dimensional filter banks are some of the frequently used speech features in speaker recognition. This speaker recognition research focuses on the usage of speech features for speaker embedding that are fitting for the speaker identification in conversational environment using a Convolutional Neural Network based approach.
  • No Thumbnail Available
    item: Conference-Full-text
    Speaker change detection for conversational speech using synthesized voice embedding
    (Information Technology Research Unit, Faculty of Information Technology, University of Moratuwa, Sri Lanka, 2019-12) Krishnathasan, M; Amalraj, CRJ; Sudantha, BH
    Speaker change detection is a prominent area of research in voice tasks for many years but used in very limited areas. Speech segmentation, feature extraction, and classification techniques are used as pipeline modules to detect speaker change in speech signals like conversations. This paper provides an approach to detect speaker changes in a conversational speech. Along with that, this paper demonstrates experiments on various sub modules and hyper parameters in the proposed pipeline.
  • No Thumbnail Available
    item: Conference-Full-text
    Speech document summarization using neural network
    (Information Technology Research Unit, Faculty of Information Technology, University of Moratuwa, Sri Lanka, 2019-12) Balasundaram, K; Amalraj, CRJ; Sudantha, BH
    Text summarization is important to reduce content overloaded. Our aim of the automatic text summarization is extract the main content from the speech in a meeting or conference for the document purpose. It is difficult to remember all the things by human. At the same time when we write about the meeting/conference there is possibility to miss the content because of some inconvenience like noises, become lazy by the long speech, getting tired and other distraction . We have proposed an automatic system which can be help to get the summarized document. At First, we translate the recorded speech in to text document using google speech recognition API. Then summarization is done using that text document.