The subjective quality ratings of noisy speech enhanced by noise reduction algorithms should be obtained according to ITU-T Recommendation P.835, which is designed to evaluate the quality along three dimensions: signal distortion, annoyance due to background noise, and overall quality. In this paper, we investigate the impact of the scale used for the assessment of the background noise dimension and show that a degradation scale and a P.835 scale yield non negligible different results. The background noise ratings as well as the overall quality ratings are impacted, the degradation scale leading to more spread and discriminating results.
Bibliographic reference. Gautier-Turbin, Valerie / Gros, Laetitia (2008): "On the perceived quality of noise reduced signals", In INTERSPEECH-2008, 2062-2065.