Skip to main content

Advertisement

Table 3 Monophonic event detection performance based on ground-truth context

From: Context-dependent sound event detection

  ACC ER A1 A30
  Global acoustic models     
Uniform event priors 32.3 85.2 10.0 21.9
Count-based event priors 36.6 84.7 12.0 25.8
  Context-dependent acoustic models     
Uniform event priors 34.7 86.9 10.9 27.0
Count-based event priors 41.1 83.4 14.8 30.2