Speaker adaptive audio-visual fusion for the open-vocabulary section of AVICAR

Keyphrases

• Audio-only 100%
• Audio-visual Fusion 100%
• Open-vocabulary 100%
• Audio 66%
• Visual Features 66%
• TIMIT 66%
• FMLLR 66%
• Neural Network 33%
• Feature Space 33%
• Regression-based Method 33%
• Word Error Rate 33%
• Automatic Speech Recognizer 33%
• Visual Speech Recognition 33%
• Forced Alignment 33%
• Convolutional Neural Network 33%
• Audio Modality 33%
• Training Targets 33%
• Speaker Adaptation 33%
• Maximum Likelihood Linear Regression 33%
• Audio-visual Systems 33%

Computer Science

• Visual Feature 100%
• Large Data Set 50%
• maximum-likelihood 50%
• Speech Recognition 50%
• Feature Space 50%
• Speech Recognizer 50%
• Word Error Rate 50%
• Convolutional Neural Network 50%