slugAutomatic Speaker Recognisation System By The Method Of Robust Formant Frequency.
Journal Title: International Journal for Research in Applied Science and Engineering Technology (IJRASET) - Year 2014, Vol 2, Issue 5
Abstract
Speaker recognition is the process of automatically recognizing who is speaking on the basis of individual information included in speech waves. This technique makes it possible to use the speaker’s voice to verify their identity and control ac cess to services such as voice dialing, banking by telephone, telephone shoppin g, database access services, information services, voice mail, security control for confidential Information areas and remote access to computers. The goal of this research is to bui ld a simple, yet complete and representative automatic speaker recognition system. Due to the limited space, we will only test our system on a mall speech database. But one can have many database files for training the system; the more files one train/teac h to the system, the more accuracy is achieved. analysis of formant tracki ng algorithms have shown that it provides accurate formant frequency estimates for both male and female speakers for a wide range in real - time noise conditions such as multiple background speakers Robust formant tracking algorithm provides mostly smooth fo rmant frequency estimates than RLS algorithm. The robust formant tracking algorithm recovers quickly after erroneous estimates to go back to tracking the actual formant frequencies in the speech signal, which is not the case with RLS algorithm. Because of this reason RLS algorithm shows noisy tracking. Information about the gender is not available with RLS algorithm. But the computation complexity of RLS algorithm is less as compared to robust formant tracking algorithm. There have been some problems identi fied with the robust formant tracker. The algorithm occasionally gives 'choppy' and oscillating formant frequency estimates. This is an undesirabl e result because the actual formant frequencies of speech normally vary slowly with time and have smooth trans itions. This problem is only encountered when the SNR is very low and occurs due to the algorithm tracking the excess energy added outside the formant frequency regions from the background noise source.
Authors and Affiliations
Abhay Kishor Tiwari, Niyati Shukla, Dr. S. K. Shrivastava
Keywords
Related Articles
- EP ID EP18046
- DOI -
- Views 285
- Downloads 11
Removal and Recovery of Phosphorus from Wastewater Using Graphene as An Adsorbent
The aim of this study is to investigate the possibility of graphene adsorbent as an alternative adsorbent for phosphorus removal from wastewater. Adsorption properties of graphene were investigated, including initial co...
The Faces of Engagement: A Method to Infer Emotions from Facial Action Units
We present a robust method to map detected facial Action Units (AUs) to six basic emotions. Automatic AU recognition is prone to errors due to illumination, tracking failures and occlusions. Hence, traditional rule base...
Advanced Surveillance Robot
In the current context of increased surveillance and security, more sophisticatedly advanced and technologically innovative surveillance systems are needed. The detection or tracking of human in highly restricted areas...
A Load Balanced Data Collection Framework for Wireless Sensor Network Using Dual Data Uploading and Polling Technique
The concept is to gather data from different mobiles in wireless sensor network. To achieve this, three layers of frame work is used. They are sensor layer, cluster head layer, mobile collector layer. This framework emp...
Information Extraction Using Enhanced N-Ary Relation Based IR Algorithm
In this paper, N-ary relation based open domain question answering system for Extraction Information from an oversized assortment of document against arbitrary questions has been presented. We proposed two algorithms to...