Skip to main content

Table 4 Objective performance: ND and NGV values using 5 and 10 training sentences, for all four gender conversions with GV enhancement (θ=2 dB)

From: Grid-based approximation for voice conversion in low resource environments

   5 Training sentences 10 Training sentences
   ND NGV ND NGV
M2M JGMM 0.76 0.6 0.74 0.55
  CGMM 0.83 0.46 0.82 0.45
  GB 0.79 0.8 0.73 0.6
M2F JGMM 0.74 0.57 0.74 0.54
  CGMM 0.83 0.45 0.84 0.46
  GB 0.76 0.73 0.73 0.68
F2M JGMM 0.77 0.63 0.75 0.69
  CGMM 0.86 0.62 0.85 0.61
  GB 0.76 0.95 0.77 1.1
F2F JGMM 0.86 0.79 0.85 0.65
  CGMM 0.91 0.63 0.89 0.6
  GB 0.95 1 0.87 0.98