Tandem Connectionist Feature Extraction for Conversational Speech Recognition thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Tandem Connectionist Feature Extraction for Conversational Speech Recognition

Published on Feb 25, 20075349 Views

Multi-Layer Perceptrons (MLPs) can be used in automatic speech recognition in many ways. A particular application of this tool over the last few years has been the Tandem approach, as described by Her

Related categories

Chapter list

Learning Long-Term Temporal Features for Conversational Speech Recognition00:00
Log-Critical Band Energies00:45
Log-Critical Band Energies01:27
What is a TRAP? 01:47
TRAPS/HATS Motivation02:41
Learn Everything in One Step?03:17
Learn in Individual Bands?03:42
One-Stage Approach03:58
2-Stage Linear->Nonlinear Approaches04:30
2-Stage MLP-Based Approaches04:47
Two Questions05:34
Experimental Setup06:14
Frame Classification07:23
Standalone Feature System08:07
ASR with Standalone Temporal Features08:52
Combination w/ PLP+09:31
Combo w/PLP Baseline Features10:15
Interpretation11:05
slide1912:09