Voice Biometrics and Voice Analytics – A primer

By Jonathon Saunders, Datapoint

We are all living in a different world today. Corporate companies and governments around the world are well informed and know exactly what security measures they want in place. Datapoint specialize in providing specific solutions to meet their business needs. Using ‘Voice’ as a process to scan and eliminate fraudulent activities. Activities such as identity theft, credit card fraud, toll fraud and money laundering. Using ‘Voice’ to assist security services such as Police, Secret Service activity, ministry of defence and Border controls. These are all customers eager to jump into this technology helping them reach their target goals.

Complimenting Biometrics is Voice Analytics. The ability to index and search on all types of audio – that could be incoming or out going from any organization. We have the ability to tie any conversation or an audio stream not only to a person, but also to search on key words, phrases irrespective of language, dialect or “slang” communication. This is extremely powerful stuff!

So let’s talk real life examples for Voice Biometrics (Finance)

Customers wishing to enquire and obtain information from Telephone banking today, have to navigate through multiple passwords and pin numbers when trying to wade through call centres. Not to mention the additional time and cost incurred by banks and call centres having call centre agents spending immense time qualifying the call.

With voice biometrics, not only can we save the bank call centres time and money, we can also ensure customers get access to their accounts efficiently and securely by using their ‘voice’ as a method of security. This in turn eliminates the need for pin numbers, passwords and memorable names and dates etc.

However one of the most compelling value propositions in this example is the ability to wipe out identity theft and fraud. The voice, your voice is unique to you, whether someone is trying to impersonate you. If your voice has aged, if your voice has changed by illness, there is no way you can bypass the voice biometric footprint.

Financial institutions, web merchandisers, Telco’s who are planning to adopt the technology have placed the financial emphasis on this client. An annual subscription of lets say anything from $1- $10 per year, enables the system to pay for its self within months.

That’s right – A solution that potentially stops white collar crime and financial fraud in its tracks – but can have an ROI within several months!

Enterprise Business drivers could be

• Validation during Telephone or Web Banking
• Money transfer
• Multiple account openings (money laundering)
• Credit card validation during transactions (over seas or above a certain financial  threshold)
• Voice Signature – replacing a wet signature (on paper)
• Validation of person within a bank branch (female in a Burca)
• Mobile Top Up
• Internal audit and risk management

So let’s talk real life examples for Voice Biometrics (state security)

Government and Security business drivers could be

• Boarder Control & E-Gate
• Stop and Search
• Illegal entry – person attempting to use additional passport to enter a country already banned from 
• Entry into Hajj or Umra

So what is the difference between Biometrics and Analytics?
Imagine having the ability to index at speed – all audio calls be it GSM, Fixed line, Satellite, Web, Video and Radio. This has not been possible until today.

Such technology is now here. Our leading solution which has the ability to index over 38000 hours of Audio (5 years worth of chat) within 24 hours. Straight from ACD or via the traditional tape systems already installed in call enters and Telco’s he added

Our solution will allow Law enforcement agencies, State Security, Telco, Call Centres to analyse all voice calls in seconds- thus, searching for key words, phrases – taking the operator/agent directly to the point in the conversation where the word/phrase was mentioned
 
It quick, accurate and can be deployed in many different vertical markets, for different objectives.

Our solution is based around phonetics. There are about 6000 languages in the world today. With around 400 phonims associated with these 6000 languages. We only use around 60 phonims in day to day chat. The beauty of our solution is that it supports
99% of all Middle East cultures and languages

Including – GCC Arabic, Modern Standard Arabic, Egyptian Arabic, South Levantine Arabic, North Levantine Arabic, Mahgebi Arabic, Farsi, Urdo, Hindi, Pashto, Chinese Mandarin, Tagalog, Russian, …

It’s also worth pointing out, Arabic slang or maybe a language spoken by a non native speaker. We will support it.

Let’s talk real life examples for Voice Analytics

There are many to choose from. However we can categorise into several groups

• Business requiring audio calls to be screened to aid in customer satisfaction and retention concerns
• Internal audit and risk management
• Government- Intelligence, Military, Police
• Legal and Corporate investigations
• Rich Media

Example One – State Security
State security, Police, intelligence is a typical end client application. Assisting the agencies which protect our lifestyles away from potential threats such as terrorism and crime.

The police have a subject under surveillance – a telephone call is monitored, indexed and searched for patterns of speech which could be viewed as anti Islamic, derogatory to the country, potential security threat or planning a crime

Once the information is indexed – it will be used to act upon either to arrest the individual, or may be as supporting evidence during trial. The key here is Speed of search and real time. If I can determine a conversation had taken place or is in the process of taking place; search the audio stream and take the investigator directly to the point in the conversation in seconds that is of interest. In turn such information can be used to quickly to arrest and prosecute.

• Real time searches for words and phrases
• Any language or slang in M.E
• Assists in protecting a culture from outside negative influence
• Demonstrates focus on bringing a level transparency to a region requiring a position on the international stage
• Any media with audio stream – This would include internet video with audio...  Again, videos which could be deemed sensitive to a society’s culture can also be indexed and tracked. Same applies to satellite, Cable, GSM, Fixed line, and Radio transmissions
 
Example Two – The Enterprise Call Center
Other less intrusive examples include Enterprise call centres wishing to monitor staff or review information quickly which could help them trend certain upturns or down turns in their business. For example

The company that is losing customers
A bank has noticed a high % of customers closing their accounts. By searching the audio in multiple languages for the words or phrases “close account” may lead them to the issue. It could well be a competitive bank is offering special incentives.

The same could apply to mobile operators, Health Care organizations... The list goes on and on.

Summary
Voice Biometrics can be adopted to organizations wishing for higher levels of security and built around existing self service applications. Target markets would be Banking, Telco’s/Mobile Operators; Web based payment platforms – working with Enterprises, Oil and Gas and also Government.

Voice Analytics – applicable to any organization wishing to search quickly on key words or phrases in real time, that could and would compromise an organization – both internally or externally. Or organizations wishing to enhance productivity and customer satisfaction by way of reviewing discussions from their contact centre’s. Target markets would be Government, Telco/Mobile, Finance, Airlines, Oil and Gas, Legal or any organization with high call volumes and rely on Contact centres for customer engagement.

For more information please contact Jonathon Saunders at jonathon.saunders@datapoint.com.