Voice can be an option for bio-metric authentication but the main question is can it provide the same level of accuracy as other bio-metric techniques provide and can we deploy it in advanced network security. We will discuss it later. First we have to know about the different types of algorithm that are used for the interpretation of the voice. These are explained as following below. These are explained as following below.
- Fixed Vocabulary Verification: In fixed vocabulary verification we are given a random set of vocabularies or we can say pool of words and for to successfully enter to the system each word in vocabulary is repeated so as to create a user model which is unique. Pool of words generally consist digits 0-9 and some different words.
- Fixed Phase Verification: It uses fixed phrase for the successful authentication and involves comparison of just two wave form. There is a tolerance b/w the two wave form and if they match within the certain limit of the tolerance then it can be assumed that this is the legitimate person and thus permission is granted. The technology used for the matching of two wave form is dynamic time warping.
- Text Independent Verification: In text independent verification you are free to speak anything, use only phrase for the authentication as now the person or the user is measured against all the other models. This method is not suitable for biometric network security as it involves continuous speech evaluation.
- Flexible Vocabulary Verification: User is able to use any word for authentication. We have to repeat series of words from lexicon that covers the whole phonemes.
How Bio-metric can be spoofed:
- Compromising with template.
- Attacking the fallback system.
- Attack the physical voice.
Article Tags :