Speech Quality Measurement
Abstract
Speech quality measurement is considered from three points of view: subjective testing; objective testing; and communicability testing. Speech quality is interpreted here in terms of user acceptability. Subjective testing is considered from the philosophical perspective of iso-preference, relative preference, and absolute-preference, with isometric and parametric test methodologies, with the results of PARM and QUART as a basis. It is felt that the best approach for future subjective testing will be a parametric approach using representative male and female talkers to cover the expected range of pitch. An automated and refined version of Voiers Diagnostic Acceptability Measure (DAM) test is an attractive option. Objective testing is considered as a possible alternative to subjective testing. Reported here is a two part experimental study of the relationship between a number of objective measures and the subjective acceptability measures available from the PARM study. In the first part of the study, controlled distortions were applied to speech samples in order to measure the resolving power of the candidate objective measures on these types of distortions. In the second part, the candidate objective measures were applied to speech samples from the same systems on which PARM tests were run, and the statistical correlation between the objective and subjective measures were studied.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 10, 1977
- Accession Number
- ADA058833
Entities
People
- A. M. Bush
- R. M. Mersereau
- R. W. Schafer
- T. P. Barnwell Iii.
Organizations
- Georgia Tech