Computer Recognition of Phonets in Speech.
Abstract
This project generated phonetic units, termed 'phonets,' from digitized speech files. The time file was converted to feature space using the Fourier Transform, and phonet occurrences were detected using Minkowski One and Two distance measures. Phonet matches were detected and ranked for each phonet compared against a template file. Phonet short-time energy was included in the output files. An algorithm was developed to partition feature space and its performance was evaluated. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1982
- Accession Number
- ADA125360
Entities
People
- Daniel L. Martin
Organizations
- Air Force Institute of Technology