March 30, 2010

Cheap & Safe Biometric Technology : Voice Verification

As more and more people and organisations depend on computers to store and save their important documents and data, there is an increasing need for security, the establishment of an end-to-end trust is highly relevant and necessary for the success of businesses and organisations relying on transactions over networks, it increases the need for more adequate user security technologies.

This article intend to introduce, a relative cost effective, easy to use and deploy biometric technology.

What is Biometric Technology ?

Biometric technologies are widely used to authenticate a person based on his/her physical and behavioral characteristics. It is more secure than the traditional user name - password or chip and pin security measures.There are two types of biometric technologies available : Behavior Based and Physical Biometric systems.

Examples of biometric technologies are : Facial Recognition, Voice Verification, Hand Geometry, Retina Recognition, Iris Recognition, etc.

How biometric technologies work ?

Instead of just verifying 'what the person knows'. Three major facts used to ensure authenticity in Biometric systems.
  • Something the person knows
  • Something the person has
  • Something the person is

Advantages and Comparison of Biometric Technologies

Biometric technologies are the most advanced methodology to authenticate a person. It helps to ensure possibly the maximum security levels to authenticate a person.

Advantages include, but not limited to,
  • Increased security
  • Eliminate problems caused by forgotten password or user-id
  • Robust and Scalable
  • Cost Effective
  • Easy to use (Exceptions Prevail)

How do I select the best one ?

The table given below compares some of the commonly used biometric technologies based on their usability and cost effectiveness.

TechnologyCost EffectiveEase of UseSpecial Hardware RequirementLow Maintenance Costs
Fingerprints
YesYesYesNo
Hand Geometry
No
Yes
Yes
No
Voice Verification
Yes
Yes
No
Yes
Iris Scanning
No
No
Yes
No
Facial Recognition
Yes
Yes
Yes
No

Table 0.1 : Comparison of Biometric Technologies

It is clear from the table that the Voice Verification Technology is a more practical solution from the comparison table (Table 0.1).

Voice Verification Technology

One of the major facts that prevent biometric technologies from being mainstream is their cost of implementation and lack of easiness to use. Many of the Biometrics technologies only make use of only two of the authentication factors discussed above. However the Voice verification outshines the other biometric technologies by using all the three factors to authenticate a person while maintaining its cost effectiveness and ease of use.


How Voice Verification works ?

Voice verification technology makes use of the distinctive characteristics of information, present in the speech samples of the speaker to verify the authenticity. Speech samples are waveforms with time on the horizontal axis and loudness on the vertical access.The voice verification system analyses the frequency content of the speech and compares characteristics such as quality, duration intensity, dynamics and pitch of the signal.

There are two types of Voice verification methodologies used, text-dependent and text-independent recognition;
  • Text-dependent version inherently increases recognition performance, but it requires a co-operative user who can remember a phrase or sentence.
  • Text-independent version does not even need the user to be aware about the system, which makes it more convenient and secure.

Enrolling a Person to a Voice Verification System is relatively simple; The user is asked to speak a word,phrase or sentence, the voice is captured and stored in the form of a voice print. This voice print is then stored into a database as voice template. Now the user has an authentication profile in the system which is equivalent to a user-name and password.

Verification of a person against the Voice Verification System is somewhat similar; where the user will repeat the word or phrase so that the system can compare the voice with the stored template to verify the identity of the person. Based on the verification system rejects or provides access to it.

There is a slight difference between how text-dependent and text-independent systems work, text-independent systems uses predetermined phrases, words or sentences and require the user to repeat the same in order to gain access. However with text-independent systems it doesn't rely on a predetermined word,phrase or sentence; the user may even be unaware of such a verification system.

Enrollment and Verification can be done by using standard microphones, telephones or Internet.

Robustness : How Reliable is it ?

This question arises all the time when voice verification is discussed, a comparison of Biometrics technique used (see appendix 1) shows that the EER of voice verification is 3.5, which is very reliable and robust.

The Voice Verification technologies are continuously improving and there are now algorithms and other techniques used to reduce the background noise to make the recording as efficient as possible.The other reason for advantage is that the user need no training; which makes it simple to handle and that makes it more reliable.

Another advantage is it doesn't need any special environment; compared to the other biometric technologies like Face Recognition which requires good lighting to work perfectly.

In addition to all this, voice verification supports vendor independent template storage; it makes it possible to upgrade or migrate to a new platform more simpler.

4.3 Security

Failure to enroll rate and Failure to accept rate are relatively low compared to the other technologies; as it has no physical barrier.

It ensure more reliability in terms of security since it has a low EER rate of 3.5 which makes it one of the most secure biometric technologies available at very low costs.

The possible attacks are comparatively low with this type of technology; Change in voice due to health, emotional states and aging can be detected with the aid of highly sophisticated algorithms, which makes it more secure. However it is pron to the Mimicking attack.

Fraud Analytics : Integration of fraud analytics is possible; it helps to achieve a new level of security strength.

Advantages of Voice Verification

The main advantage that voice recognition provides is, it can be used remotely over telephone or using Internet. Even-though voice verification is vulnerable to certain types of mimicking attacks, it is widely used in Voice dialing, Tele-Banking, remote access of computers, security control for confidential information,etc .It's low cost maintainability, ease of use and low deployment cost makes it stand out among the other major biometric technologies.

The table above compares the different available biometric technologies; it is clearly visible that the EER rate of Voice Verification is lower; it is third major technology in terms of EER. Lower EER rate indicates the better security.
  • Remote access is possible, Which can provide an easy and comfortable way to authenticate customers without them physically come down to the organisations location.
  • Enable the organisation to provide access to a service with higher level of security irrespective of the customer location.
  • Very convenient to use; no more complex passwords or user names or Long alpha-numeric strings for authentication .
  • Increased level of trust; which will help in increasing the status quo of organisation.
  • Perfect Component of multi factor biometric authentication (something the user knows, something the user is).
Cost Effectiveness & Simplicity
  • Low implementation costs
  • Low maintenance costs
  • Reduce in investment, in existing technologies like password recovery,etc
  • Easy to use
  • Higher Return On Investment
  • No Special Training needed for the users; Saves initial training cost.
Each new additional security needs more monetary investment; in addition to this the amount of money spent for the it infrastructure management associated with the new technology will increase the expenditure. However the voice verification helps to reduce the initial cost and maintaining costs.

A case study by Jamison consultants in 2009 reveal that depending upon the industry,It spends on resetting lost passwords of employee itself can account for 10-60% of IT calls at a cost of $10-45 per call.With the voice verification technology we can either remove the need for password managed system or even we can integrate a voice verification system to automate the password reset system, which will help to save thousands of dollars per year.

Hardware and related expenditure can be reduced by deploying voice verification; most of the other biometric technologies including face recognition requires more amount of investment in special hardware; However voice verification can even be deployed with standard microphones; a quick Froogle search reveals that there are microphones available from a price starting at $4 per piece. Bulk buying can reduce this cost even further; resulting in very cheap implementation costs. There are Open Source software solutions like Asterisk VoIP which can be adapted and customized for specific needs with very low cost for over-phone voice verification.

Compared to the competing technologies Voice Verification provides a Higher Return On Investment Rate.

A 2009 survey by Lieberman research group shows that 69 % of users are willing to use voice verification as a biometric security technique.


How easy to Choose, Deploy and Maintain Voice Verification

Choosing Technology

There are a wide variety of products available in the market; which suits different needs and requirements of the organizations or customers. It makes the selection process much easier. Nuance is one of the leading vendors in the market.

However there even alternative opensource platforms which can be adapted to custom needs; such as CMU Sphinx by Carnegie Mellon University .

The availability of wide range of hardware solutions for cheaper costs is one of the major advantage of choosing voice verification technology.

Deployment & Maintenance

The most easy to deploy biometric technology is Voice Verification; as it does not use any special hardware it is very easy to deploy and maintain. Microphone arrays can be implemented with ease to use it for more demanding environments where more users can verify themselves against the system at once.

A remote scenario of phone based voice verification system can easily be managed using a VOIP solution. Which will enable the organization to save a lot of investment amount. There are industry standard opensource VOIP Solutions like Asteriskt available in the market; which are free to download and customise to fit the need of the organisaiton.

The users need no additional training to use this technology as some other biometric technologies require training to start with the technology; the users some times even need not know about the existence of technology. This makes it more convenient to manage and maintain.

Since it is possible to use universally acceptable voice template databases; the migration to a newer more advanced technology needs no additional enrollments or end-user involvement; which makes it more convenient for future migrations and up-gradations.

Conclusion ! Oh Yeah It might work for me ! (???)

Now you have a basic idea of what voice verification is, and what are it's advantages. Hope this will help you consider Voice Verification, while you are looking for a cost effective and secure biometric technology for your organization.

--

References

[1] Ziv Barzilay. (2007). Voice Biometrics as a Natural and Cost-Effective Method of Authentication. CellMax Systems Ltd White Paper. 0 (1), 16.

[2] G. Doddington, “Speaker Recognition-Identifying People by their Voices,” Proc. IEEE, Vol. 73, November 1985.

[3] John D. Woodward, Jr., Nicholas M Orlans, and Peter T. Higgins (2003). Biometrics. New York: McGraw Hill Osborne.

[4] P. Jonathon,A. Martin, C.L. Wilson, M. Przybocki. (2000). An Introduction to Evaluating Biometrics Systems. Available: http://www.frvt.org/DLs/FERET7.pdf. Last accessed 3 Mar 2010.

[5] Lazo, Manuel; Sanfeliu, Alberto (Eds.) (2004). Progress in pattern recognition, image analysis and applications. Havana, Cuba: Springer. 703.

[6] Yee Wah Lau, Michael Wagner, Dat Tran. (2004). Vulnerability of Speaker Verification to Voice Mimicking. Proceedings of 2004 International Symposium on Intelligent Multimedia, Video and Speech Processing. 0 (0), 145-148.

[7] Quest Biometrics. (2005). Advantages of Biometrics. Available: http://www.questbiometrics.com/advantages-of-biometrics.html. Last accessed 7 Mar 2010.

[8] Feustel, T.C. Velius, G.A.. ( 27-30 Nov 1989). Voice-based security: identity verification over telephone lines. Global Telecommunications Conference, 1989, and Exhibition. Communications Technology for the 1990s and Beyond. GLOBECOM '89., IEEE. 1 (1), 212 - 216.

[9] Nancy Jamison. (2009). Your voice is your key. Available: http://www.jamison-consulting.com/pdf/WP9-036NVoiceAuthentication-VoiceisKey8-2009.pdf. Last accessed 6 Mar 2010.

[11] Google Labs. (2010). Froogle Search for Microphones.Available: http://www.google.co.uk/products?q=microphone&scoring=p. Last accessed 6 Mar 2010.

[12] N.L. Clarke, S.M. Furnell. (2007). Advanced user authentication for mobile devices. Computer & Security. 26 (1), 109-119.

[13] Wikipedia. (2003). Biometrics. Available: http://en.wikipedia.org/wiki/Biometric#Performance. Last accessed 19 Mar 2010.

[14] Lieberman Research Group. (2009). UNISYS Security Index: United States 25 September 2009 (2H’09). Available: http://www.unisyssecurityindex.com/resources/reports/US%20Security%20Index%20Oct%2009.pdf. Last accessed 19 Mar 2010.

[15] Andrew Shaw. (1997). Voice Verification - Authenticating Remote Users Over the Telephone. Network Security. 1997 Issue 8 (1), 16-20.

[16] Marcia Gulesian. (2007). Speech Authentication Strategies, Risk Mitigation, and Business Metrics. Available: http://www.developer.com/security/article.php/11580_3684921_2/Speech-Authentication-Strategies-Risk-Mitigation-and-Business-Metrics.htm. Last accessed 19 Mar 2010.

[17] ACA Research. (2005). 2005 Speech Recognition Customer Satisfaction Survey.Available: http://www.freeformdynamics.com/misc/docs/Nuance_Customer_Sat_ExecSumm.pdf. Last accessed 19 Mar 2010.

[18] Jonathan Kent. (2005). Malaysia car thieves steal finger.Available: http://news.bbc.co.uk/1/hi/world/asia-pacific/4396831.stm. Last accessed 20 Mar 2010.

[19] National Biometric Security Project. (2005). Biometric Technology Application Manual. Available: http://www.nationalbiometric.org/btamvol1update.pdf. Last accessed 19 Mar 2010.

[20] Indy Yu,Tian He. (2004). A User Dependent Voice Verification System. Massachusets Institute of Technology, US, Theses. 1 (1), 112.

[21] OAM Portal. (2010). Voice Authentication. Available: http://oamportal.fdvs.com/fdvssite/products/voice-biometrics.aspx. Last accessed 19 Mar 2010.

[22] Nuance. (2010). Nuance Website Home Page. Available: http://www.nuance.com/. Last accessed 20 Mar 2010.

[23] Carnegie Mellon University. (2009). CMU Sphinx - Speech Recognition Tool Kit. Available: http://cmusphinx.sourceforge.net/. Last accessed 20 Mar 2010.

[22] Digium. Inc. (2010). Asterisk : Opensource VoIP telephony solutions. Available: http://www.asterisk.org/. Last accessed 20 Mar 2010.


No comments: