Academic Journals Database
Disseminating quality controlled scientific knowledge

Survey Of Privacy-Preserving Audio Representations With Speaker Diarization

Author(s): S.Sathyapriya M.phil1 , A.Indhumathi , M.Phil , Ph.D

Journal: International Journal of Computer Trends and Technology
ISSN 2231-2803

Volume: 4;
Issue: 9;
Start page: 3059;
Date: 2013;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Speaker diarization | Deep neural network (DNN) | Linear Prediction (LP) residual | MFCC features.

Investigation of the audio features for speaker diarization and verification with privacy in multiple conversations by considering low linguistic information becomes major important issues in Audio Representations. The Major aim of the speaker verification is the process of finding speaker change points in an audio stream. The subsequent aim at group together speech segments on the foundation of speaker characteristics. For this speaker diarization identifying the important features with privacy Linear Prediction (LP) residual is the one of the main model to represent the important audio features with efficient bandwidth calculation and subsignal features also. Previous approaches to privacy sensitive features have paid attention to moreover reinterpreting easy, frame-level heuristics for estimate talking activity in the conversation only with either log files .In this study focus on finding the audio features low linguistic information using deep neural architecture for deriving privacy-sensitive audio features. To admiration this concept of privacy, features might be stored beginning which neither a comprehensible verbal communication nor the lexical contented can be reconstructed. Using this move toward to extract privacy- sensitive features and then to apply diarization. Deep neural network (DNN) with a blockage structural design support audio features with MFCC using the diarization. Finally study the previous methods and deep forward neural network shows better audio feature identification than previous linguistic feature based information.
Save time & money - Smart Internet Solutions      Why do you need a reservation system?