Term Project
Close-Set, Text-Dependent, Twelve(12) people, Speaker Identification Using ANN
(Artificial Neural Networks)
JAY DESAI KUANG-TAO CHIAO
Introduction
OverviewClosed Set/Open SetText Dependent/Text IndependentSpeaker Identification/Speaker
Verification
System Architecture
Block Diagram
Some Plots we obtained
Short-time Energy
The 4 vowelsShort time energyLog plot
Frame Extraction
3 frames/vowel168 Cepstral Coeff.
Password
/u/ /i/ /æ/ /a/Why the choice of password?Vowel PlaneThe Phoneticians vowel trapezium
Linear Predictive Coding
Why LP analysis?Feature ExtractionComputational aspectsLPC Cepstrum
Artificial Neural NetworksArtificial Neural Networks
wki
θk
yk
nk
Back Propagation
Potential Applications
Meetings, Conferences, Conversations
Law enforcementSecurity applicationHuman-Machine InterfaceGender recognitionOthers
Scope of Improvement
RobustnessAdditive NoiseCo-channel InterferenceIncreasing the number of users