LREC2004_Mixer_Poster

Course: PAPERS 2004, Fall 2008
School: UPenn
Rating:
 
 
 
 
 

Document Preview

Mixer The Corpus of Multilingual, Multichannel Speaker Recognition Data Christopher Cieri1, Joseph P. Campbell2, Hirotaka Nakasone3, David Miller1, Kevin Walker1 1 University of Pennsylvania, Linguistic Data Consortium, Philadelphia, PA, USA 2 MIT Lincoln Laboratory, Lexington, MA, USA 3 Federal Bureau of Investigation, Quantico, VA, USA ccieri@ldc.upenn.edu, j.campbell@ieee.org, hnakasone@fbiacademy.edu,...

Register Now

Unformatted Document Excerpt

Coursehero >> Pennsylvania >> UPenn >> PAPERS 2004

Course Hero has millions of student submitted documents similar to the one
below including study guides, practice problems, reference materials, practice exams, textbook help and tutor support.

Course Hero has millions of student submitted documents similar to the one below including study guides, practice problems, reference materials, practice exams, textbook help and tutor support.
Mixer The Corpus of Multilingual, Multichannel Speaker Recognition Data Christopher Cieri1, Joseph P. Campbell2, Hirotaka Nakasone3, David Miller1, Kevin Walker1 1 University of Pennsylvania, Linguistic Data Consortium, Philadelphia, PA, USA 2 MIT Lincoln Laboratory, Lexington, MA, USA 3 Federal Bureau of Investigation, Quantico, VA, USA ccieri@ldc.upenn.edu, j.campbell@ieee.org, hnakasone@fbiacademy.edu, damiller@ldc.upenn.edu walkerk@ldc.upenn.edu Forensic Automatic Speaker Recognition Fisher Platform Behavior 350 300 Callers by Calls Made Minimum Requirements text-independent Callers 250 channel-independent 200 New requirements capable of handling multiple languages including bilingual speakers 150 100 50 Plan multilingual, multi-channel collection dissemination to research sites system performance improvement system performance evaluation 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 36 Calls Completed Multichannel System Laptop with two firewire hard drives Multichannel interface Sensors 0. Wireline telephone call goes to robot operator platform, not to MOTU) 20 Calls by Language Corpus Design Speakers 100 100 100 100 100 100 100 100 100 100 100 50 4 Calls 10 Arabic Mandarin Unique Russian Handset Spanish XC & TR Extended Data Xchannel Extended Data Extended Data Extended Data Extended Data Extended Data 2. Cell-phone style ear-bud in-line lapel mic Motorola Earbud Handsfree, SYN8390, $12 3. Over the ear miniboom mic Jabra EarWrap Headset Radio Shack #43-1914 $30 4. Courtroom mic Shure MX418S Supercardioid Gooseneck Mic $185, mounts to furniture 5. Conference room mic (table top boundary mic) Crown SoundGrabber II pressure-zone mic $70 6. Distant mic (e.g., courtroom mic across the room) Phase I Arabic Mandarin Russian Spanish II Research Supports English Phase speaker recognition research with emphasis on forensicstyle problems: telephone conversations channel independence language independence & bilingualism transcript reading Generalities Audio Technica Pro 45 Cardioid Condenser Hanging Mic $91 7. Studio mic (placed near talker) Audio Technica AT3035 Cardioid Condenser $200 (not including stand) 8. PC-style stand mic Radio Shack Desktop Mic with Noise Canceling #33-3031 $27 9. Microcassette mic Olympus Pearlcorder S725 (no mic monitoring capability) $66 calls are 6 minutes in duration subjects speak to each other assigned topics change daily robot operator logs ANI unique handsets encouraged subjects indicate phone and handset type Components Core: 600 subjects*10 calls Extended: 100 subjects continue to 20 calls Multingual 100 subjects * 4+ of their calls in Arabic 100 subjects * 4+ of their calls in Mandarin 100 subjects * 4+ of their calls in Russian 100 subjects * 4+ of their calls in Spanish Multi-channel Data: 100 subjects * 4 of their calls via multi-channel device Transcript Readin...

Find millions of documents on Course Hero - Study Guides, Lecture Notes, Reference Materials, Practice Exams and more. Course Hero has millions of course specific materials providing students with the best way to expand their education.

Below is a small sample set of documents:

UPenn - PAPERS - 2004
recent activities in resource creation and distribution and the development of tools and standards Christopher Cieri, Mark Liberman{ccieri,myl}@ldc.upenn.edu University of Pennsylvania Linguistic Data Consortium and Department of Linguistics 3600 Ma
UPenn - PAPERS - 2004
Dialectal Arabic Telephone Speech Corpus: Principles, Tool design, and Transcription ConventionsMohamed Maamouri, Tim Buckwalter, Christopher CieriLinguistic Data Consortium University of Pennsylvania maamouri@ldc.upenn.edu, timbuck2@ldc.upenn.edu,
UPenn - PAPERS - 2003
Robust Sociolinguistic Methodology:Tools, Data and Best PracticesChristopher Cieri, Stephanie Strassel {ccieri, strassel}@ldc.upenn.edu University of Pennsylvania Linguistic Data Consortium and Department of Linguistics 3600 Market Street, Philadel
UPenn - PAPERS - 2002
Resources for Arabic Natural Language ProcessingMohamed Maamouri, Christopher Cieri {maamouri,ccieri}@ldc.upenn.eduUniversity of Pennsylvania Linguistic Data Consortium and Department of Linguistics www.ldc.upenn.edus International Symposium on P
UPenn - PHYS - 1
Lecture 2 IntroRunway Problem, pg. 1Runway Prob., pg. 2Lect. Quest. 1: ramp prob.Lect. Quest. 1: ramp contd.Ramp Prob. Soln., pg. 1Ramp prob. Soln: time c-gLect. Quest. 2: vel. vs. timeLect. Quest. 2: choose graphsLect. Quest. 3: re
UPenn - PHYS - 1
Lecture 3 IntroDefinition of DynamicsNewtons Laws of MotionTypes of ForcesSpring ForcesNormal and friction forcesTensionNormal Forces explainedThe WeightWeight is Normal ForceFind Weight in an ElevatorTension ForcesConnect Que
UPenn - PHYS - 1
Lecture 4 IntroRadioactive DecayRadioactive Decay: statisticalRadioactive Decay: equationRadioactive Decay: exponentialHalf-Life definedNuclear Medicine definedConnect Quest: O15 minimumLect. Quest. 1: max. C leftLect. Quest. 2: 1/1
UPenn - PHYS - 1
Lecture 5 IntroIntro. To Uniform Circ. MotionDefinition of Uniform circ. Mot.Definition of tangential vel.Connect Quest.: Tangential Acc.Definition of RadiansRadians and degreesKinematics of circ. motionDefinition of period and ang. V
UPenn - PHYS - 1
Lecture 6 Intro PageOscillatory Motion - IntroConnection to Circular MotionOsc. And circles - use ScreenshotPhysical picture of diff. Eqs.Robert Hooke and his LawHookes Law and SpringsConnect Question: Eq. For Osc.Ang. Frequency for r
UPenn - PHYS - 1
Lecture 7 - Intro SlideConversations About ConservationDescartes Bold AssertionNewtons Bolder AssertionNewtonian SynthesisWhy A New Second Law?The Incomplete F = maNew Definition of MotionDefining the SystemDefining the System - col
UPenn - PHYS - 1
Lecture 8 IntroMore ConservationThere is More to Conserve!Kinetic Energy and ForcesThe Definition of WorkRefining the Work DefinitionStatement of Work-Energy TheoremExample 1: Work done on boxConnect Question 1: Centripetal Motion and
UPenn - PHYS - 1
Lecture 9 - Rotational MotionThe Universe and RotationRewind of Angular MotionRigid Body Angular MotionRigid Body Angular MotionRigid Body Angular DynamicsRadial Acceleration ReviewedWhy Use Angular Variables?Connect Question: calcula
UPenn - PHYS - 1
Lecture 10 - IntroductionRotations and The Universe continuedTranslation and RotationRotational and Linear Equations for motionMore Rotational and Linear EquationsPure Rolling MotionConnecting Rotation to TranslationWhat Is No Slipping?
UPenn - PHYS - 10
Lecture 10 - IntroductionRotations and The Universe continuedTranslation and RotationRotational and Linear Equations for motionMore Rotational and Linear EquationsPure Rolling MotionConnecting Rotation to TranslationWhat Is No Slipping?
UPenn - PHYS - 1
Lecture 11 - IntroCentral Forces and GravitationOrbital ViewSidereal Not SynodicWhy Arent Orbits Simple?The Law of Universal GravitationWhy an Inverse Square Law?Remarks on the TheoryMore Remarks.And One Last RemarkWhy g is Consta
UPenn - PHYS - 11
Lecture 11 - IntroCentral Forces and GravitationOrbital ViewSidereal Not SynodicWhy Arent Orbits Simple?The Law of Universal GravitationWhy an Inverse Square Law?Remarks on the TheoryMore Remarks.And One Last RemarkWhy g is Consta
UPenn - CML - 3
Neighborhood Information SystemCreated by Cartographic Modeling Lab University of Pennsylvania Funded by William Penn Foundation The Pew Charitable Trusts University of PennsylvaniaCore Concept Work with City agencies to create an integrated parc
UPenn - LDC - 2004
Dialectal Arabic Telephone Speech Corpus: Principles, Tool design, and Transcription ConventionsMohamed Maamouri, Tim Buckwalter, Christopher CieriLinguistic Data Consortium University of Pennsylvania maamouri@ldc.upenn.edu, timbuck2@ldc.upenn.edu,
UPenn - SSC - 2005
CONFIDENTIALEducation for the Examination vs. Education for Holistic Development-The Transformation of Teacher Beliefs and Practices in Rural Northwest ChinaDocument Date Tanja Carmel Sargent University of Pennsylvania, Graduate School of Educat
UPenn - CIS - 03
DISE: A Programmable Macro Engine for Customizing ApplicationsMarc Corliss, E Lewis, Amir Roth University of PennsylvaniaCorliss, Lewis, + Roth ISCA-30Overview Application customization functions (ACFs) E.g., safety-checking, debugging, decomp
UPenn - CIS - 05
Low Overhead Debugging DISEwithMarc L. Corliss E Christopher Lewis Amir Roth Department of Computer and Information Science University of PennsylvaniaOverview Goal: Low overhead interactive debugging Solution: Implement efficient debugging pr
UPenn - CIS - 04
Using DISE to Protect Return Addresses from AttackMarc L. Corliss, E Christopher Lewis, Amir Roth University of PennsylvaniaOverview Prevent stack-smashing attacks Old approach, new implementation Dynamic Instruction Stream Editing (DISE)DISE
UPenn - WHARTONSBD - 100
Congr atu la tio ns to Phila delp hia 100 !Philadelphia has the third most companies in the 2003 Inc. 500, with 21, behind Washington, D.C. (41) & Boston (23)Outline Describe selection process andcriteria Describe the winners Recognize entre
UPenn - CSE - 100
Lecture II Sept 8, 2003Lawrence Lessigs Four Modalities of Regulation & Cyberspace9/8/03CSE 100 - Lecture 21Lessig: Constitutions for Cyberspace By constitution, I dont mean a legal text. Rather, as the British understand when they speak o
UPenn - CSE - 100
Lecture 3 Sept 11, 2003Stories of Cyberspace: Special properties & the role of architecture9/11/03CSE 100 - Lecture 31Four constraints on behavior (recap) Law regulates by sanctions imposed ex post If youre under age you cant sm
UPenn - CSE - 100
Privacy of Communications: Snail Mail to TelephonesPrivacy: Snail mail 1710 Postal service in British colonies created. Opening letters forbidden without authorization from secretary of state. 1878 Supreme Court rules (Ex Parte Jackson) governm
UPenn - CSE - 100
Pen Registers & Packets9/29/03CSE 100 - Info Tech & Its Impact on Society1Privacy: Pen RegistersUnited States v. Miller (1976) No constitutional right to business records held by third party. Smith v. Maryland (1979) Given US v Miller, pen
UPenn - CSE - 100
Pen Registers & Packets II10/02/03CSE 100 - Info Tech & Its Impact on Society1Fisher suspends tactic in fighting child porn Philadelphia Inquirer 9/10/2003Pennsylvania Attorney General Michael Fisher yesterday agreed to halt his behind-the-
UPenn - CSE - 100
Data Privacy: Social Security Numbers10/9/03CSE 100 - Info Tech & Its Impact on SocietyAuthorized Uses of Social Security Numbers 1943 Federal agencies use SSN exclusively for employees. 1961 Civil Service Commission uses SSN as an e
UPenn - CSE - 100
The following slides are adapted from:Introduction to Web Security By Adam Cain2nd WWW Workshop, South Korea , 11/1/95and from a tutorial at the website of PrivacyExchange http:/www.privacyexchange.org/Symmetric-key Cryptography Data encrypt
UPenn - CSE - 100
Introduction to Intellectual PropertySources include: Introduction to Intellectual Property Rights, U.S .Dept of State www.usinfo.state.gov/products/pubs/inteprp and Copyright Basics, U.S. Copyright Office www.copyright.gov/circs/circ1.html10/30/2
UPenn - CSE - 100
Understanding the Digital Millennium Copyright Act (DMCA)Much of the presentation here adapted from: Jessica Mitman, Digital Copyright, Prometheus Books, 200111/6/2003CSE 100 Info Technology & Its Impact on SocietyOr rather, the ideas behind
UPenn - CSE - 100
MP-3 Compression: How it works11/11/03CSE 100 Info Technology & Its Impact on Society1WHAT IS MP3?"MP3 is a compression form . (which) tries to eliminate the frequencies which the human ear is unable to hear. As a result the MP3 compression
UPenn - CBIL - 05
Crackingthepromotercode: Identifyingregulatorymodules fortissuespecifictranscripts.Chris Stoeckert, Ph.D. Center for Bioinformatics & Dept. of Genetics University of Pennsylvania School of Medicine Nov.15,2005 UniversityofNebraskaMedicalCenter Wh
UPenn - CBIL - 04
ComputationalAnalysisof TissueSpecificity:Decoding PromotersChrisStoeckert,Ph.D. CenterforBioinformatics&Dept.ofGenetics UniversityofPennsylvania Nov.17,2004 DepartmentofPhysiologySeminarSeries UniversityofKentuckyWhatisthecodefordeterminingwhe
UPenn - CBIL - 404
AnnotatingMicroarray DatawiththeMGED OntologyP.L.Whetzel,A.Pizarro,E.Manduchi,J.Liu,H.He, G.Grant,M.Mailman,C.Stoeckert CenterforBioinformatics UniversityofPennsylvaniaNCICenterforBioinformatics April15,2004Science298:601604,2002Science298:597
UPenn - CBIL - 2
DifferentialExpressionandFalse DiscoveryRateRevisitingthePrincetonStemCellData GPBAWorkshopOct.15,2003GeneExpressionAnalysisofHematopoieticStemCells andCommittedProgenitorsHongxianHe1,GregoryGrant1,LyleUngar2,NataliaB.Ivanova3,IhorR.Lemischka
UPenn - CBIL - 03
GeneratingUsefulInformationin Toxicogenomics:FocusedEfforts: Feb.6,2003,TheNationalAcademiesChrisStoeckert,Ph.D. Centerfor Bioinformatics Universityof PennsylvaniaMicroarrayStandardsMicroarrayStandardsAreNeededTo FacilitateMovingDataAround
UPenn - C - 2
SharingMicroarray ExperimentKnowledgeChipstoHitsOct.28,2002 ChrisStoeckert,Ph.D. Dept.ofGenetics&CenterforBioinformatics UniversityofPennsylvania Nature,October3,2002http:/plasmodb.org/David Roos, Jessie Kissinger, Bindu Gajria, Martin Fraunhol
UPenn - CBIL - 02
Integrated Data Systems for Genomic AnalysisGenomicsandBioinformaticsforthe AdvancementofClinicalSciences ThomasJeffersonUniversity,Oct.14,2002ChrisStoeckert,Ph.D. Dept.ofGenetics&CenterforBioinformatics UniversityofPennsylvania Plasmodiumgenomi
UPenn - CML - 8
SUMS CategoriesData Elements, INatality:Adequate Prenatal Care Adequate Prenatal Care, percent Births to Teen Mother Infant Deaths Infant Deaths, percent Low Birth Weight Babies Low Birth Weight Babies, percent Pre-term Births Pre-term Births, pe
UPenn - GROUP - 040802
PENN Community Project SUG PresentationApril 8, 2002PENN Community Project Agenda Overview Background Phase PhaseI Objectives I Deliverables of Benefits Summary Futures Wrap-Up2PENN Community Project OverviewCollect and manage the
UPenn - CML - 6
Brownfields Inventory Pilot ProjectPrepared for the Pennsylvania Environmental Council by the Cartographic Modeling Lab University of Pennsylvania June 2000Brownfields Inventory Pilot ProjectPennsylvania Environmental CouncilCartographic M
UPenn - SOMAPPS - 00014
TranslationalResearchABasicScience Perspective Translationalresearch:translatesbasicscience discoveriesintoclinicalapplications,and/orusesclinical observationstogeneratenewresearchtopics. Focusisontheintegrationofactivitiesfrombenchto bedside. Tra
UPenn - SOUTHASIAC - 2
The Basics of HinduismPatel StyleIntroduction to Siddhartha Academic English IVThe Vedas Massive holy book of Hinduism Written in Vedic, an ancient form of Sanskrit Collection of four different booksRig Veda Most important Veda Collectio
Villanova - CSC - 02
Chapter 2Data Design and ImplementationDefinitionsAtomic or primitive type A data type whose elements are single, non-decomposable data items Composite type A data type whose elements are composed of multiple data items Structured composite type
Villanova - CSC - 2053
1052SeriesforGraphicsGraphics, Applets Graphical User Interfaces1 CLH GUI slidesOutline - Chapter 2Graphics Applets Drawing Shapes2 CLH GUI slidesIntroduction to GraphicsThe last few sections of each chapter of the textbook focus on gra
Villanova - CSC - 2053
CSC 2053New from 5.01AutoBoxing2AutoBoxingBefore J2SE 5.0, working with primitive types required the repetitive work of converting between the primitive types and the wrapper classes. The new autoboxing feature in J2SE 5.0 handles convers
Villanova - CSC - 2053
Java I/OWriting and Reading Objects to File SerializationStoring Objects/Structures in Files Many programs need to save information between program runs Alternately, we may want one program to save information for later use by another program
Villanova - CSC - 2053
Chapter 2Data Design and ImplementationDefinitionsAtomic or primitive type A data type whose elements are single, non-decomposable data items Composite type A data type whose elements are composed of multiple data items Structured composite type
Villanova - CSC - 2053
Design Patterns and RefactoringCSC 2053Jan 21, 2009BuzzwordsDesign Patterns describe the higher-level organization of solutions to common problemsDesign Patterns are a current hot topic in O-O designUML is a diagramming language design
Villanova - CSC - 2053
Binary Search TreesIn Chapter 9, binary search trees are used to implement bags and sets. This presentation illustrates how another data type called a dictionary is implemented with binary search trees.1The Dictionary Data TypeA diction
Villanova - CSC - 2053
HEAPS & PRIORITY QUEUESArray and Tree implementations1Priority queueA stack is first in, last out A queue is first in, first out A priority queue is least-first-outThe smallest element is the first one removed The definition of smallest is up
Villanova - CSC - 2053
CSC 2053 -TREESAVL & BALANCED TREES1Balanced TreesThe advantage of balanced trees is that we can perform most operation in time proportional to O(log n) But what is a balanced tree?A balanced binary tree is a binary tree where all leave
Villanova - CSC - 2053
QuicksortThis is probably the most popular sorting algorithm.\ It was invented by the English Scientist C.A.R. Hoare It is popular becauseit works well in a variety of situations it is a good general purpose algorithms 1QuicksortThe d
Villanova - CSC - 2053
MERGESORT -Csc 2053 SORTING1CSC 20531Stable vs. Non-Stable SortsWe frequently use sorting methods for items with multiple keys Sometimes we need to apply the sorting with different keysFor instance we want to sort a list of people b
Villanova - CSC - 2053
Introduction to GraphsSeveral problems in the real-world can be formulated as objects and connections between them.e.g.What is the fastest way to get from city A to city B? What is the cheapest way to get from city A to city B? The above can
Villanova - CSC - 2053
Hash TablesThe crucial disadvantage for avoiding direct access structures such as arrays is that we need to allocate in advance the size of this structure We tend to overestimate the its size and end up with a very sparse structure We tend to t
Columbia - WWW1 - 2003
DesignDeveloping Software on LinuxTim Marriott Mel Nicholson ICCAD 2003Introduction Why Linux? Porting to Linux Developing on Linux 2003 Synopsys, Inc. (2)Why Linux? The question should be Why the PC? Inexpensive Performance What is wr
Columbia - WWW1 - 2003
grid Mat rix Tec hno log yGrids in EDA Software DevelopmentTom Grotton Cadence Design Systems, Inc. Director, IT Server Farm InitiativeCADENCE CONFIDENTIAL 1Corporate Needs Need more computing capacity Products are becoming more complex and ta
Columbia - WWW1 - 2001
ESUIF: An Open Esterel CompilerStephen A. Edwards Department of Computer Science Columbia University, New York www.cs.columbia.edu/~sedwardsCopyright 2001 Stephen A. Edwards All rights reservedNot Another One My research agenda is to push Este