What is Speaker Verification
Speaker verification, also known as speaker recognition, can be defined as process of authenticating ones claimed identity by comparing their voice and previously recorded or computed speech characteristics (voiceprint, which is as unique as finger print ). While speech recognition is widely used in IVR, speaker verification or recognition is different than speech recognition and is relatively recent addition to numerous usage of IVR systems.

Speaker verification is one of the recent and advanced IVR feature which is increasingly being used some IVR applications where personal identification is very important, for example, Banking, attendance registration etc.

How it can be used in IVR ?

As the name suggests, speaker verification is essentially used for authentication a IVR caller to manage his account. Since now, IVR systems have been authenticating its caller with ‘password’ or PIN ( Personal Identification Number) by pressing telephone keys or DTMF or even speech recognition.

Normally, while using speaker recognition for user authentication, IVR caller is asked to speak a random phrase which is compared with caller’s existing voice print. This is a quicker process for user authentication.

Advantages of Speaker Verification

The age-old IVR user authentication method of asking caller to enter a 4 digit ( or 5 digit) PIN is slow as well as it needs to changed periodically. But with usage of speaker recognition, authentication process is faster and IVR may ask a caller to speak any phrase randomly without user having to remember PIN or keep changing it periodically.

Another method of caller verification was using ANI (Automatic Number Identification) or Caller ID. But though detection of caller ID is fool proof, but this kind of user authentication may not be fool proof as anybody could dial from user’s phone number. But speaker recognition with voice print does not suffer from these drawbacks. Speaker recognition along with speech recognition may provide almost fool prof user authentication.

How it helps usability of IVR

With speaker recognition, customer details could be retrieved quickly from CRM ( Customer Resource Management) and many routine questions to a caller could be avoided. Caller would be happier not having to repeat his information each time he calls to the IVR.

Since it authenticates a caller much faster, it can provider any service quicker to the caller and usability of the IVR increases.

Many people may feel frustrated having to enter PIN or change it periodically for security reasons. But with speaker recognition, a caller just has to speak a given phrase and can authenticate herself. Many people. specially the one, who use call center or IVR rarely, may forget PIN. With speaker recognition, user does not need to remember any PIN or change the PIN periodically.

CRBT (Caller Ring Back Tone) is a very popular IVR application used in Telecom Industry. It is known as “Hello Tunes”, “Caller Tunes” in India. CRBT is a fairly simple IVR system which has a smart call routing and a IVR application which plays one audio file, a music as per choice by the CRBT subscriber. CRBT is one of the highest revenue earner for Telecom companies in voice based value added service (VAS) all around the world.

CRBT can also be used as very useful, targeted advertising tools where the IVR software will play an ad jingle instead of a number for chart topper, or a old hit. The ad jingle to be played can be chosen by the following criteria:
1. Caller Id : Since IVR application will know both the caller number and called party number, it can choose ad jingle based on the caller ID. For example, if the caller id is an land-line number, a ad jingle related to land line facilities should be played. When caller ID is a Airtel land line number, why not play a internet facilities available on Airtel land-line number ?
2. Area of Caller ID : Area of caller can be decided based on the caller id. In case the caller ID is a mobile number, first 4 digits of the caller ID will determine the area of the caller. And if the caller id is a land line number, the first 3 digits which is a STD (Subscriber Trunk Dialing) code which denotes an area. So, using this information, the ad jingle can be chosen.
3. Telecom Operator Wise : From the caller id and the called party ID (DNI, Dialed Number Identification), Telecom operator can be found out. Advertisement can be targeted by Telecom Operator too. For example, all BSNL caller would be played about 3G availability on BSNL network!
4. Integration with Billing Server : If IVR Server or CRBT Server is connected to billing server, than high paying customers, prepaid customer, post paid customer can be identified and accordingly ad jingles can be targeted.

Advantages of CRBT advertisement :
Advertising on CRBT is new. But it is obvious that there are certain advantages of CRBT advertising. It can be very targeted advertising. The advertiser would know exactly how many times ad jingles has been played and to whom. He would be able calculate his conversion rates.

I think CRBT advertisement would pick up in India soon. There are people who are experimenting on it. But we are still to see wide scale adoption of this advertisement platform.

High-End digital CTI boards for a voice portal is not cheap! In fact some of them cost as much as a Honda Civic car in India. So, choosing a right CTI Board for your Voice Portal may not be easy, that is until you read this article .

So, how to choose a CTI Board for your voice portal application ? There are few parameters using which one should choose CTI Board for Voice Portal:-

1. Number of Expected Simultaneous Calls
Most important factor to consider! If you expect simultaneous calls to be not more than four, buy any four channel analog card. But than, most probably it will not remain a voice portal! Ideally, if one should choose a CTI board with 50% more capacity than the expected number of glutenous calls. if some one expects 10-15 calls, he should choose one 30 port Digital CTI board.

2. Time to connect a call
Time taken for connecting a call when an user call is important! It depends of the interface the CTI board supports. For example, CAS R2MF connects much slower than SS7 or ISDN PRI. One will hear “teent teent teent” sound after dialing the Voice Portal before connection. Normally quick and instant connection to Voice Portal is desired.

3. Interface With Switch Support
The type of interfacing with switch of Telecom Provider is important parameter to consider. Many Telecom Service Provider may not have CAS or ISDN protocols on their switch or they may not be willing to provide that connectivity to Voice Portal operators. So, one must consider about available protocol with Telecom Provider before deciding on a CTI board. Interfacing with various protocols is responsible time to connect of calls. SS7 cards are costlier than ISDN PRI or CAS R2MF boards, connects calls almost instantly. And some CTI Manufacturer like Dialogic does not have a combined board which has both Voice Processing and SS7 Signaling so far! This kind of arrangement has its own advantage over combined CTI boards. For example, one can have add more Media Processing (Voice Processing) Boards while keeping one signaling board in one system. Also combined (Signaling and Media Processing) CTI boards do not come in low density configuration, they normally have 8 E1 or more. 4E1 meaning at least 120 simultaneous calls!

[ad#image-large]

3. Scalability and IVR Software
This is another aspect one should not ignore. If the Voice Portal becomes popular, which is the main aim by any Voice Portal Operator, so the CTI Boards should have scalability feature where increasing number of simultaneous calls should not make existing hardware redundant, or make IVR Software obsolete. Separate boards for signaling and media processing is recommended for scalability. IVR Software should be designed such a way that, it can handle more voice resources without having to re-develop the IVR Software.

4. Support for the CTI Board
Availability of development support as well as quick 24 hours support for any problem in the CTI hardware is a must! Every minutes of downtime of Voice Portal, the operator loses revenue! So, one should choose a CTI vendor which can provide quick support or quick replacement of CTI hardware in case of CTI board failure! Also, CTI Vendor should have active forum to discuss various issues and should have good number of third party developers!

5. Cost
Cost is very important, well for everybody! But, one should not compromise of quality of CTI cards for cost.

My Favourite and Recommended CTI hardware:

1. For 30 to 120 simultaneous Calls.
Start with Dialogic D/300 JCT or DMV1200 card. They have ISDN, R2MF protocol, good support, quite tested CTI board. Many developers have written IVR Software on them! Dialogic has good support, active forum, large number of developers. But since they are the biggest CTI vendor, they may not provide you with a demo card for development of your Voice Portal!
These boards can be even used with Dialogic SS7 Signaling boards.

2. For 240 or more simultaneous calls
1. Synway SHD-240D-CT/PCI
Good Value for money, good support. Built in Signaling and media processing. Cheaper than Dialogic equivalent.
2. Dialogic SPCI4/SPCI2 and DNI2410
Awesome combination, great scalability, very good support and large number of developers as well as very active forum. Cost is almost double that Synway though!