My Voice is My
Passport: Verify Me
March 12, 2012
About the Speakers
   Dan Miller (Founder, Senior Analyst)
       Founded Opus Research (1985)
       Analyst at IDC/Link, The Kelsey Group, Zelos Group
       Industry experience: Atari, Time-Warner, PacTel (AT&T)
       Expertise: local, DA, speech integration with Web, mobility
        and enterprise software infrastructure
       Coverage Areas: Conversational Commerce, Internet2Go,
        Biometrics
   Isaac Chapa (VP Information Systems/Operations,
    CSID)
       Joined CSID 2006, overseeing ID Theft Platforms/Solutions
       Sr. Engineer, Grande Communications
       Designed and integrated DCM/VOIP Switches, Billing
        Platforms, SONET and FTTH, HFC networks

                                                                      2
Why We’re Here
 Talk about voice biometrics
 Share some ideas on stronger authentication for
  mobile transactions
 Get feedback as prospective
  users/developers/implementers
 Describe some “real world” use cases, business cases
  and demand drivers




                                                         3
Voice Biometrics & Speaker Verification
 Voice Biometrics is a technology
    Captures an utterance from a live caller
    Compares it to previously stored “voiceprint”
    Produces a score
 Speaker Verification is an application
    Employs a biometric engine plus business logic
    Enrolls customers by obtaining voice prints
    Compares live utterances to voice prints to produce a
    “pass” or “fail” responses




                                                             4
Speaker Verification Components
 Core Verification Engine
    Receives voice sample (“utterance”); compares it to a voiceprint
     (“template”)
    Confirms who said it
 Core Recognition Engine
    Compares utterance to ASR grammar
    Determines what was said
 Business Logic
    Decides if the caller passes or fails
    Dictates required “next steps”




                                                                        5
What is Voice Print?
Physical Characteristics         Behavioral Characteristics
The unique physical traits of    The harmonic and resonant
the individual’s vocal tract,    frequencies, such as accents, the
such as shape and size.          speed of your speech, and how
                                 words are pronounced and
                                 emphasized.




      Voiceprint - Together these physiological and
      behavioral factors combine to produce unique voice
      patterns for every individual


                                                                     6
Text Dependent vs. Text Independent
 Applications that require a specific pass phrase are Text
  Dependent
    Require training
    Customarily involve enrollment
 Text Independent applications can use any utterance
    Simplify enrollment
    Support “conversational authentication”




                                                              7
Why Now?




           8
Fraud Protection Requirements
   Multifactor
       Mandated in more use cases
       Includes “something you are”
   Multimodal
       Because “the customer is always on”
       Embraces social networks and multiple sign-ons
   Mobile
       Approaching 6 billion subscribers
       Mobile devices are becoming virtual assistants




                                                         9
+ 1 = Momentum
 Passwords getting more difficult
    Multiple digits and special characters
    Frequently updated
    Fragmented across sites (and IDs)
 User authentication vital
    To access multiple sites, domains and devices
    For more activities, transactions and interactions
    “Open” approaches only as strong as weakest link




                                                          10
Entering 3rd Generation
                                                    Password replacement
                                                    Leverage device id
                             Voiceprint on identity Random phrase liveness
“My voice is my              claim                       3rd Generation
password”
                             Leverage KV & ANI/CLI
“ 0 1 2 3 4 5 6 7 8 9”                                  Enhanced ID&V
                             Random word liveness
Random digit liveness                                Secure Mobile Access
                                  2nd   Generation      Voice Signatures
     1st Generation                                         Embedded
                                                       -Internet via OOB
                                 Enhanced ID&V              Verification
                                                     - Mobile multi-media
  IVR PIN replacement            Multi-factor Auth        Convenience
    Password Reset                Automation
  Emphasis on Security

2000                  2005                    2010           2015
 Source: Nuance Communications

                                                                            11
Estimated Revenues




                     12
Results: Registered Voiceprints
                 In Millions




                                  13
This is My Wallet




                    14
This is My Wallet on Phone




                             15
Mobile Commerce is Exploding
 Mobile transactions
    Will reach $670 billion by 2015
    Up from $240 billion in 2011
 Global in nature
    East Asia and China
    Western Europe
    North America

    represent 75% gross transaction value.
    (Juniper Research)




                                             16
But Inherently Insecure
 At the device level
    OSes have no security shell
    Personal info (including PINs) stored as text
 At the network level
    Encryption is the exception
    Prone to keystroke logging, Bluetooth sniffing and the like


                What about authenticating users?




                                                                   17
What Are We Protecting
 Integrity
 Confidentiality
 Availability of Data
    Loosely coupled from infrastructure
 Secure applications and runtime environments
 The critical focus of security shifts:
    From owning everything to owning nothing
    From “Where are you from?” to “Who are you?”
       • Identity, credential, and access management
    From “Internal vs. External” to “Distrust everyone equally”
 Need strong authentication independent from current
  form factors


                                                                   18
What Are We Using
 Usually a four digit number.
    There's only 10,000 possible combinations
    Four character, alpha only, password has more than
     45,000 possibilities
    Alphanumeric and there's more than a million and a half
 Fast computers can crack these in less than a
  second (and often don’t have to)




                                                               19
Today’s Requirements
 “Layered”
    To apply appropriate level of security for
    risk profile
 Multi-Factor
    To augment PINs or PWD
 Device-oriented
    Complex device identification
    considered more secure
    (per 2011 “guidance” from FFIEC)




                                                  20
Lead To These Solutions
 Treating mobile phones as “non-
  traditional endpoints”
 Popular solutions:
    One Time Passwords – using SMS text
    Knowledge-based Authentication –
     using non-public info
    “A Biometric” – fingerprint, face
     recognition, iris scans…and voice!




                                           21
Before You’d Try These




                         22
You Should Think About These
 User Authentication
 Device Activation
 Transaction Authorization
 Mobile Signatures
 Password Reset
 ID Proofing




                               23
Superior Factor for Phones
 Works on all phones
 Includes both physical and behavioral attributes
 Physical Characteristics        Behavioral Characteristics
 The unique physical traits of   The harmonic and resonant
 the individual’s vocal tract,   frequencies, such as accents,
 such as shape and size.         the speed of your speech, and
                                 how words are pronounced and
                                 emphasized.




                                                                 24
On Par With Biometric Alternatives
 Error rates are “acceptable”
 Registration is relatively easy
 No special equipment needed for authentication
 Solutions integrate with or augment existing security
  infrastructure




                                                          25
Has Surprising Acceptance
 In contact centers
    8.5 million voice prints registered
    ROI justified shaving minutes from authentication practices
    + fraud reduction
 For remote and mobile workers
    For Password Reset
    Secure access to VPN
    Strong authentication for conference calls




                                                                   26
Applications & Use Cases
 Personalized, trusted customer care
 Proof of life
 Mobile payment authorization
 Device activation, “Wake Up”
 Enterprise VPN access control
 Password reset
 Anonymous authentication




                                        27
But Real Security Comes With
 Layering multiple factors
    Like gestures
    Location
    Motion detection
    Out-of-band authentication
 And leveraging existing infrastructure
    For compliance
    As a go-faster
    To support Natural Language Interactions




                                                28
Thank you.
Dan Miller
Dmiller@opusresearch.net
@dnm54 on Twitter

Isaac Chapa
ichapa@csid.com




                           29

Sxsw ppt voice-1

  • 1.
    My Voice isMy Passport: Verify Me March 12, 2012
  • 2.
    About the Speakers  Dan Miller (Founder, Senior Analyst)  Founded Opus Research (1985)  Analyst at IDC/Link, The Kelsey Group, Zelos Group  Industry experience: Atari, Time-Warner, PacTel (AT&T)  Expertise: local, DA, speech integration with Web, mobility and enterprise software infrastructure  Coverage Areas: Conversational Commerce, Internet2Go, Biometrics  Isaac Chapa (VP Information Systems/Operations, CSID)  Joined CSID 2006, overseeing ID Theft Platforms/Solutions  Sr. Engineer, Grande Communications  Designed and integrated DCM/VOIP Switches, Billing Platforms, SONET and FTTH, HFC networks 2
  • 3.
    Why We’re Here Talk about voice biometrics  Share some ideas on stronger authentication for mobile transactions  Get feedback as prospective users/developers/implementers  Describe some “real world” use cases, business cases and demand drivers 3
  • 4.
    Voice Biometrics &Speaker Verification  Voice Biometrics is a technology  Captures an utterance from a live caller  Compares it to previously stored “voiceprint”  Produces a score  Speaker Verification is an application  Employs a biometric engine plus business logic  Enrolls customers by obtaining voice prints  Compares live utterances to voice prints to produce a “pass” or “fail” responses 4
  • 5.
    Speaker Verification Components Core Verification Engine  Receives voice sample (“utterance”); compares it to a voiceprint (“template”)  Confirms who said it  Core Recognition Engine  Compares utterance to ASR grammar  Determines what was said  Business Logic  Decides if the caller passes or fails  Dictates required “next steps” 5
  • 6.
    What is VoicePrint? Physical Characteristics Behavioral Characteristics The unique physical traits of The harmonic and resonant the individual’s vocal tract, frequencies, such as accents, the such as shape and size. speed of your speech, and how words are pronounced and emphasized. Voiceprint - Together these physiological and behavioral factors combine to produce unique voice patterns for every individual 6
  • 7.
    Text Dependent vs.Text Independent  Applications that require a specific pass phrase are Text Dependent  Require training  Customarily involve enrollment  Text Independent applications can use any utterance  Simplify enrollment  Support “conversational authentication” 7
  • 8.
  • 9.
    Fraud Protection Requirements  Multifactor  Mandated in more use cases  Includes “something you are”  Multimodal  Because “the customer is always on”  Embraces social networks and multiple sign-ons  Mobile  Approaching 6 billion subscribers  Mobile devices are becoming virtual assistants 9
  • 10.
    + 1 =Momentum  Passwords getting more difficult  Multiple digits and special characters  Frequently updated  Fragmented across sites (and IDs)  User authentication vital  To access multiple sites, domains and devices  For more activities, transactions and interactions  “Open” approaches only as strong as weakest link 10
  • 11.
    Entering 3rd Generation Password replacement Leverage device id Voiceprint on identity Random phrase liveness “My voice is my claim 3rd Generation password” Leverage KV & ANI/CLI “ 0 1 2 3 4 5 6 7 8 9” Enhanced ID&V Random word liveness Random digit liveness Secure Mobile Access 2nd Generation Voice Signatures 1st Generation Embedded -Internet via OOB Enhanced ID&V Verification - Mobile multi-media IVR PIN replacement Multi-factor Auth Convenience Password Reset Automation Emphasis on Security 2000 2005 2010 2015 Source: Nuance Communications 11
  • 12.
  • 13.
  • 14.
    This is MyWallet 14
  • 15.
    This is MyWallet on Phone 15
  • 16.
    Mobile Commerce isExploding  Mobile transactions  Will reach $670 billion by 2015  Up from $240 billion in 2011  Global in nature  East Asia and China  Western Europe  North America represent 75% gross transaction value. (Juniper Research) 16
  • 17.
    But Inherently Insecure At the device level  OSes have no security shell  Personal info (including PINs) stored as text  At the network level  Encryption is the exception  Prone to keystroke logging, Bluetooth sniffing and the like What about authenticating users? 17
  • 18.
    What Are WeProtecting  Integrity  Confidentiality  Availability of Data  Loosely coupled from infrastructure  Secure applications and runtime environments  The critical focus of security shifts:  From owning everything to owning nothing  From “Where are you from?” to “Who are you?” • Identity, credential, and access management  From “Internal vs. External” to “Distrust everyone equally”  Need strong authentication independent from current form factors 18
  • 19.
    What Are WeUsing  Usually a four digit number.  There's only 10,000 possible combinations  Four character, alpha only, password has more than 45,000 possibilities  Alphanumeric and there's more than a million and a half  Fast computers can crack these in less than a second (and often don’t have to) 19
  • 20.
    Today’s Requirements  “Layered”  To apply appropriate level of security for risk profile  Multi-Factor  To augment PINs or PWD  Device-oriented  Complex device identification considered more secure (per 2011 “guidance” from FFIEC) 20
  • 21.
    Lead To TheseSolutions  Treating mobile phones as “non- traditional endpoints”  Popular solutions:  One Time Passwords – using SMS text  Knowledge-based Authentication – using non-public info  “A Biometric” – fingerprint, face recognition, iris scans…and voice! 21
  • 22.
  • 23.
    You Should ThinkAbout These  User Authentication  Device Activation  Transaction Authorization  Mobile Signatures  Password Reset  ID Proofing 23
  • 24.
    Superior Factor forPhones  Works on all phones  Includes both physical and behavioral attributes Physical Characteristics Behavioral Characteristics The unique physical traits of The harmonic and resonant the individual’s vocal tract, frequencies, such as accents, such as shape and size. the speed of your speech, and how words are pronounced and emphasized. 24
  • 25.
    On Par WithBiometric Alternatives  Error rates are “acceptable”  Registration is relatively easy  No special equipment needed for authentication  Solutions integrate with or augment existing security infrastructure 25
  • 26.
    Has Surprising Acceptance In contact centers  8.5 million voice prints registered  ROI justified shaving minutes from authentication practices + fraud reduction  For remote and mobile workers  For Password Reset  Secure access to VPN  Strong authentication for conference calls 26
  • 27.
    Applications & UseCases  Personalized, trusted customer care  Proof of life  Mobile payment authorization  Device activation, “Wake Up”  Enterprise VPN access control  Password reset  Anonymous authentication 27
  • 28.
    But Real SecurityComes With  Layering multiple factors  Like gestures  Location  Motion detection  Out-of-band authentication  And leveraging existing infrastructure  For compliance  As a go-faster  To support Natural Language Interactions 28
  • 29.
    Thank you. Dan Miller Dmiller@opusresearch.net @dnm54on Twitter Isaac Chapa ichapa@csid.com 29