A study of speech compression using analog time domain sampling techniques
Abstract
Fourier transform methods were used to establish the correspondence between the quality of the reconstructed compressed speech waveforms and the subjective recognition of compressed speech. For maximum intelligibility, it was shown that the sampling has to be synchronous with the pitch periodicity of the voiced speech waveform. The result of the two psychoacoustic experiments indicated that: (1) the interruption frequency should be equal to the pitch frequency of the voice waveform for the optimum recognition of compression speech, and (2) smoothing of the discontinuities significantly improves the recognition of compressed speech. The optimum smoothing parameters, window width and characteristics function are also obtained from this study.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- 1976
- Bibcode:
- 1976PhDT........73B
- Keywords:
-
- Algorithms;
- Analog Simulation;
- Speech;
- Data Smoothing;
- Fourier Transformation;
- Mathematical Models;
- Minicomputers;
- Parameterization;
- Prototypes;
- Speech Recognition;
- Time Functions;
- Waveforms;
- Communications and Radar