Comparing a magnitude estimation technique and a pair comparisons technique for use in assessing quality of text-to-speech synthesis systems
Chaslav V. Pavlovic, Christel Sorin, Jean Pierre Roumiguiere, Jean-Pierre Lucas · The Journal of the Acoustical Society of America · 1989
This study was designed to compare a magnitude estimation procedure and pair comparisons procedure for use in scaling quality of text-to-speech synthesis systems. To this goal, the psychophysical scale values that result from the magnitude estimation procedure were compared to the values obtained from a pair comparisons procedure. Four different synthesis systems and three different prosody rules in all possible combinations were scaled using a magnitude estimation technique. To obtain the pair comparisons scale values, the results reported by Lucas et al. [CNET report NT/LAA/TCS/375, Tom I, Lannion, France (1988)] for the same systems were used. They were reanalyzed for this purpose. The data indicate a good general agreement between the results of the two experiments. Thus the two methods seem to validate each other. The relative efficiency of the magnitude estimation technique is greater than that of the pair comparisons technique.