The Use of Voice in Biometric Authentication

Voice control has been a used on a large scale in improving security as there can be certain situations which they perfectly fit. Since majority of people across the world use English, the adversary takes advantage of this fact. In this proposal, I think that if exceptional phrases or words in different languages are included, difficulty level for the adversary will increase.

Introduction

Let us consider is a banking center, where a large number of people call for account related issues. When the account holder is calling, they need to answer the security questions. If answers match data-base, the system automatically accords the caller. There is a possibility that any adversary has all the data and they can also pretend to be the authentic account holder. Data theft can also occur if someone physically enters the building without proper identification and that can be detrimental. The traditional system of sliding or sweeping employee card is not that effective as id-cards can be forged very easily.

Process

Online/Telephonic Banking:

• There has to be a password phrase which is generated by the system depending on what languages one is familiar. Whenever they call, even before asking the guarantee questions, the structure should query them about this phrase.
• This phrase can’t be replicated with the password given for logging into the account.
• They can use a phrase or a word that can be concatenated with a random keyword that is generated by the system.
• If someone knows Spanish, then they can speak a word in Spanish and then the system will generate a random keyword which will be in another language.
• All the customers will be given three attempts to say the correct phrase. If they fail, then the system will automatically disconnect the call.

Employee access:

• All the employee will have a unique employee id. Firms do have a habit of giving old employee number to a new candidate which may lead to breach of security.
• The system will give them randomly selected phrases and they can’t select their own phrase.
• Each employee is given a different phrase.
• The employees will be given three attempts to say the phrase.
• If the employee fails the authentication, they would have to get a new phrase.
• Employees should have access to only the floor in which they work.

Advantages:

• Words from multiple languages installed in the system will increase the variety of phrases that can be used.
• The combination of word can be more than two. However, the customer or the employee may not prefer that scenario.
• It can protect against Ventriloquists who can mimic the speech of people.
• Automatically disconnecting the call after three failed attempts or not allowing the employee to enter the building is also a key feature as it helps in making the system more secure.

Disadvantages:

• There are certain words that sound similar or have same spelling in the languages. If we take them into consideration, it might flop to comprehend the attack. That’s why it’s very important to have a list of such words. The system has to make sure that such words can’t be used as a voice password phrase.
• Maintenance of database would be a huge challenge as attackers find out advanced techniques to hack the data stored. They can try different combinations and figure out the key phrase for each account.
• If someone is speaking in lower volume, the system may not be able to follow the phrase.
• It is preferred if the person knows more than one language.

Future Scope

It is very hard to answer which biometric would be most preferred but including different type of measures can make voice recognition more effective. Speech pattern might be different depending on the age group to which a person belongs. One of the concerns that rise is that the words are given to person without knowing about their knowledge and wisdom. They may not be able to pronounce or understand the word and can make repeated mistakes. However, if the system can take into consideration their knowledge and expertise, it will be easier for people.

Conclusion

We can conclude that voice is a strong biometric authentication if certain predicaments are changed. If the numeral for words is large, different phrases can be generated. I don’t think that mass would be willing to use this as they might think that time is a constraint. They can’t be all day on phone verifying themselves by trying different phrases. Even the bank employees won’t be happy with this as it’s a continuous authentication process. With time we can presume that biometrics will be an asset for our systems.