Skip to main content

Table 3 Comparison between the proposed method for degradation detection and the baseline system for pathological voice and normal running speech

From: A CNN-based approach to identification of degradations in speech signals

Detectors Pathological voice Normal running speech
  Baseline Proposed Baseline Proposed
Noise 0.96 ±0.00 1.00 ±0.00 0.71 ±0.00 0.95 ±0.01
Distortion 0.90 ±0.01 0.99 ±0.00 0.83 ±0.01 1.00 ±0.00
Reverberation 0.75 ±0.00 0.93 ±0.01 0.84 ±0.01 0.99 ±0.00
  1. The results are in the form mean AUC ±95% confidence interval, and the bold-faced numbers represent the best performance