Skip to main content

Advertisement

(a) Scenario 1: "neutral versus emotional ASR engine"

From: On the Impact of Children's Emotional Speech on Acoustic and Language Models

    M   E   A
Baseline system 43.6 61.3 64.9
Adapted systems    
Acoustic models 43.1 74.8 73.5
Linguistic models 49.3 67.0 68.5
Both 47.4 76.5 75.3