1 Scope
2
References
3
Abbreviations
4
Definitions
5
Conventions
6
Summary of objective measurement procedure
7
Source speech material preparation
7.1 Real voices
7.2 Artificial voices
8
Selection of experimental parameters
9
Calculation of objective quality
9.1 Global initializations
9.1.1 Time alignment
9.1.2 Global scaling
9.1.3 Global calibration
9.2 Time-frequency mapping
9.2.1 Windowing
9.2.2 Sampled Spectral Power Density (SPD)
9.3 Frequency warping and filtering
9.3.1 Sampled Pitch Power Density
9.3.2 Local scaling
9.3.3 Telephone-band filtering
9.3.4 Hoth noise
9.4 Intensity warping
9.5 Cognitive modelling
9.5.1 Loudness scaling
9.5.2 Sampled noise disturbance density
9.5.3 Asymmetry processing
9.5.4 Noise disturbance including silent
interval processing
10 Transformation from the objective
quality scale to the subjective quality scale
10.1 Mean opinion scores
10.2 Equivalent-Q values
11 Analysis of results
Appendix I – Contents of floppy diskette accompanying Recommendation P.861
I.1 Introduction
I.2 \test directory
Appendix II – Objective quality measurement of telephone-band (300-3400 Hz)
speech codecs using Measuring Normalizing Blocks (MNBs)
II.1 Introduction
II.2 Computation of the objective measure
II.2.1
Input-output specifications
II.2.2 Time delay
II.2.3 Signal preparation
II.2.4 Transformation to frequency domain
II.2.5 Frame selection
II.2.6
Perceived loudness approximation
II.2.7 Frequency Measuring Normalizing Block
(FMNB)
II.2.8 Computing Time Measuring Normalizing
Blocks
II.2.9 Linear combination of measurements for
MNB structure