Speaker Verification

1
Speaker Verification

• Seth McNeill
• 18 April 2003
• Presentation for EEL6586 Automatic Speech
  Processing

2
Outlineof things to come

• 1) Who am I?
• 2) Verification vs. ID
• 3) Features
• 4) Modeling
• 5) My Project

2/12
3
Who am I?

• First Year Graduate Student
• First Year at UF
• From SE Washington State

3/12
4
(No Transcript)
5
Speaker VerificationvsSpeaker ID

• Verification Are you an imposter?
• ID Which of N speakers are you?

4/12
6
Features

• Mel-Cepstrum, D Cepstrum, Cepstral Mean
  Subtraction
• Glottal Flow Derivative
• Liljencrants-Fant (LF) model
• Sub-Cepstrum

96
95
74
5/12
7
Gaussian Mixture Model (GMM)

• Loses Temporal Data
• Person vs. Background
• Person vs. Threshold

6/12
8
GMM (continued)
7/12
9
My Project

• C Implementation
• Energy Based Endpoint Detection
• Mel-Cepstrum, D Cepstrum Coefficients
• Single Window with Nearest Neighbor
• Muliple Window with GMM

8/12
10
Current Progress

• Data Capture from Sound Card
• Endpoint Detection

9/12
11
Demo
10/12
12
Future Progress

• Feature Extraction
• Modeling
• Motion Detection
• Text-to-Speech
• Visual Verification

11/12
13
Speaker Verification
Questions or Comments?
12/12
14
Another GMM Video