Workshop on Speaker biometrics and VoiceXML 3.0 — Agenda

General information:

Note that some presentations will contain material that is relevant to multiple sessions. Thus, material may be presented in an earlier session than the one in which it is discussed.
Presentation time is strictly limited to 30 minutes to ensure sufficient time for questions and discussion at the end of each session.
During the discussion, new issues that are identified will be added to the "discussion list" which will be reviewed during the summarization session.
Attendees will be asked to write detailed descriptions of issues during the conference. These descriptions will be inserted into the conference minutes.

Thursday, 5 March 2009

8:45-9:15 Session1: Welcome

Moderator:: Jim Larson
Scribe:: Jim Larson

Welcome and logistics — Jordan Cohen, SRI

Welcome from the W3C — Kazuyuki Ashimura, W3C [ Slides ]

9:15-10:30 Session2: Background

Moderator:: Kaz Ashimura
Scribe:: Dan Burnett

Overview of applications — Judith Markowitz, J. Markowitz, Consultants [ Paper / Slides ]

Topics to discuss:

Who is the audience for this standard?
How is SIV being used now?
Who is using it?

Question and answer:

Should VoiceXML functions be made available to non-IVR applications?
Should work on an SIV specification for VoiceXML 3.0 be constructed with a broader set of deployments in mind? If so, how can we define the limits of the scope of that work?

SIV in VoiceXML 3.0 — Ken Rehor, Cisco Systems [ Slides (TBD) ]

Topics to discuss:

What does VoiceXML.2.0 and VoiceXML 2.1 look like?
What are its limitations?
What does VoiceXML 3.0 look like?
What has the VBWG already discussed in with regard to the SIV specification?

Question and answer

10:30-11:00 BREAK

11:00-12:00 Session3: SIV Use Cases

Moderator:: Judith Markowitz
Scribe:: Bruce Balentine

Speaker Verification in a Multi-Vendor Environment — Ross Summerfield, Centrelink [ Paper / Slides ]

Position Paper: VoiceXML Speaker Bio,etrics Workshop — Alfred Tom, General Motors [ Paper / Slides (TBD) ]

Topics to discuss:

What kind of functionality does this application need in an SIV specification of VoiceXML?
What are the biggest challenges in implementing/adopting SIV in this application?
Does this application use text-dependent and text-independent technology?
What kind of support is needed for multiple engines/modalities?

Question and answer

12:00-12:30 Session4: Users

Moderator:: Kaz Ashimura
Scribe:: Ross Summerfield

Speaker Identification & Verification — Bruce Balentine, EIG [ Paper / Slides (TBD) ]

Topics to discuss:

What the "design philosophy" could/should be?
What kind of mechanism would be useful for integrating uncertain recognition results from multiple engines/modalities/devices? Would EMMA and/or SCXML be useful?
What specific features/functions does the SIV specification need in order to support the uncertainty principle?
What specific features/functions do we need to avoid putting into the SIV specification? If they are core features of SIV how can we modify them to support the uncertainty principle?
How much of the "sniffing around" is the responsibility of the enterprise vs. the developer?

Question and answer

12:30-2:00 LUNCH

2:00-2:45 Session5: Audio Formats

Moderator:: Ken Rehor
Scribe:: Chuck Johnson

A Standard Audio Encapsulation Method — Homayoon Beigi, Recognition Technologies [ Paper / Slides ]

Topics to discuss:

What criteria need to be satisfied in order for an audio-format to be included in the minimal list of standard audio formats for SIV?
What standard audio-formats should be included in the minimal set of formats that can be used with the SIV specification?

Question and answer:

Are there conditions/needs that would not be satisfied by this minimal list of open standards?
Are there audio formats that are not interoperable or convertible to this short list?

2:45-3:30 Session6: Relationship of SIV Specification with MRCP V2

Moderator:: Judith Markowitz
Scribe:: JP Shipherd

Dan Burnett, Voxeo [ Slides ]

Topics to discuss:

What is MRCP V2?
What are the major ways in which MRCP V2 differs from SOAP?
Advantages and disadvantages of each as the standard way to integrate IVR and SIV engine?
How well do these standards support buffered, stored, and streaming audio?

Question and answer:

Should the SIV specification contain an interface to MRCP V2? To SOAP? Both?

3:30-4:00 BREAK

4:00-5:30 Session7: Architecture and functionality

Moderator:: Jim Larson
Scribe:: Ken Rehor, JP Shipherd

A possible SIV architecture discussed in Menlo Park (Menlo Park Model) — Bruce Balentine, EIG [ Slides ]

Proposal of an SIV Architecture and Requirements — Ingmar Kliche, Deutsche Telekom [ Paper / Slides ]

Configuration and Management of Speaker Verification Systems — Chuck Johnson, iBiometrics [ Paper / Slides ])

((Alcatel-Lucent (not attend) [ Paper ]))

((PerSay (not attend) [ Paper ]))

Topics to discuss:

What functionality is included in each proposed SIV specification?
Does this SIV specification support stored, buffered, and streaming audio?
Does the SIV specification or the application decide whether to accept or reject the user?
Who manage the voice model database? SIV application developers, system managers, security professionals?
Who is responsible for enrollment? SIV application developers, system managers, security professionals?
Where exactly do you draw the line between what should be performed by the voice browser and what should be performed by the application server? For example what should handle "delete voice model" or "fetch voice model?"
Do specific technology approaches (e.g., HMM, neural networks) need different requirements? How independent is the technology from the architecture?
Does this architecture support continuous verification?
Does this architecture support watchlists? A watchlist is a prespecified list of users who should not be allowed to use the system.
What is the structure of the API and how does it pass parameters?

Question and answer:

What functionality should be included in the SIV specification?
How should the API work?
Can non-VoiceXML applications access VoiceXML functions?

Note. Scribes for March 5 submit their session summaries to Kazuyuki Ashimura

5:30 Day1 ends

7:00- DINNER

Friday, March 6 2009

8:30-9:20 Session8: Security and identity management

Moderator:: Cathy Tilton
Scribe:: Ken Rehor

[Quick reminder that there is security terminology in the glossary]

Security, Identity Management, and Risk Management — Valene Skerpac, iBiometrics [ Paper / Slides ]

Topics to discuss:

How do we take privacy, security, and risk management into consideration when we develop an SIV specification?

Question and answer

9:20-10:00 Session9: Relationship with CBEFF and INCITS 456

Moderator:: Valene Skerpac
Scribe:: Homayoon Beigi

Common Biometric Exchange Format Framework (CBEFF) — Catherine Tilton, Daon [ Slides ]

Topics to discuss:

What is CBEFF?
How does it work?

INCITS 456 – Data Interchange format for SIV — Judith Markowitz, J. Markowitz, Consultants [ Slides ]

Topics to discuss:

What is INCITS 456?
What is the relationship between EMMA and INCITS 456?

Question and answer:

Should SIV specification support INCITS 456?
Are there any other CBEFF formats that SIV specification should support?

10:00-10:05 SHORT BREAK

10:05-10:30 Session10: Data format for Multi-modal and Multi-Factor Applications

Moderator:: Dan Burnett
Scribe:: Ingmar Kliche

[Review of terminology — multi-factor and multi-modal including multi-SIV engine]

Extensible Multi-Modal Annotation Markup Language (EMMA) — Jim Larson, Intervoice [ Slides ]

Topics to discuss:

What is EMMA?
Why is it useful for SIV?

Question and answer:

Are there other notations that would compete with EMMA?
Should SIV specifications support EMMA or something like EMMA?
Should there be a fusion/decision engine? Should it support EMMA?

10:30-11:30 Session11: Relationship with the BIAS Project and BIOAPI

Moderator:: Ken Rehor
Scribe:: Kaz Ashimura

The Biometric Identity Assurance Services (BIAS) Project and BioAPI — Catherine Tilton, Daon [ Paper / Slides on BioAPI / Slides on BIAS ]

Topics to discuss:

What is the BIAS project?
What is BioAPI?
How does SIV fit into BioAPI and BIAS?
Should we consider using BioAPI as represented by the BIAS committee as the SIV specification? If not, should we consider using BioAPI as a guide or source for development of that specification?

Question and answer:

Should BioAPI be used as the SIV specification?
Should the SIV specification contain an interface to BioAPI as a way of supporting multi-biometric applications?

11:30-12:00 LUNCH

12:00-12:45 Session12: Proposed SIV Architecture

Moderator:: Jim Larson
Scribe:: Judith Markowitz

VoiceXML 3 SIV Extensions: Pros and Cons — JP Shipherd [ Slides ]

12:45-1:00 SHORT BREAK

1:00-2:00 Session13: SUMMARY: WHAT HAVE WE DECIDED?

Moderator:: Jim Larson
Scribe:: Judith Markowitz

Summarization — all participants

Topics to discuss:

Each participant will be asked to identify the key take-away for them from the meeting.
The list of action items will be reviewed.
What are the next steps we should take?

Goodbye — Kazuyuki Ashimura, W3C

Note. Scribes for March 6 submit their session summaries to Kazuyuki Ashimura

2:00 Workshop ends

The Call for Participation, the Logistics and the Presentation Guideline are also available on the W3C Web server.

Judith Markowitz, Ken Rehor and Kazuyuki Ashimura, Workshop co-Chairs

$Id: agenda.html,v 1.11 2007/01/09 14:06:52 ashimura Exp $