DID YOU KNOW? DTIC has over 3.5 million final reports on DoD funded research, development, test, and evaluation activities available to our registered users. Click
HERE to register or log in.
Accession Number:
ADA151898
Title:
Processing Speech for Analysis Using Optical Fourier Techniques.
Descriptive Note:
Master's thesis,
Corporate Author:
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH SCHOOL OF ENGINEERING
Report Date:
1984-12-01
Pagination or Media Count:
109.0
Abstract:
In this thesis a system for displaying speech as a two dimensional video image is presented. The speech is pre-processed by compressing its dynamic range and filtering to emphasize frequencies above 500 hz. Blanking and sync pulses are inserted to put the signal in standard video format, and every other field is blanked to prevent interference between fields in the interlaced display. Two dimensional variation is achieved by modulating the baseband audio signal up in the spectrum near a multiple of the video scan rate. The relationship between input frequency and pattern angle of the display is derived, and it is shown that the set of frequencies near a multiple of the video scan rate have points in the spatial frequency domain which lie in a straight line at a distance from the origin proportional to the scan rate multiple. Two modulation frequencies are selected to display in the spatial frequencies domain the location of the first and second formant peaks. The two modulated signals are mixed with the baseband audio and displayed simultaneously in a single image. The images are digitized and an optical Fourier transform is simulated on the computer by creating the image which would appear in the Fourier transform plane. Entire words are processed by assembling individual frames on video tape. Additional keywords Speech Recognition, Phonemes, Spatial Filtering. Author
Distribution Statement:
APPROVED FOR PUBLIC RELEASE
