SPEECH-TO-CODE CONVERTER STUDY
Abstract
A study was conducted to determine the parameters necessary for the identification of words spoken in isolation. Segmentation procedures were developed which isolated the minimal unit (a syllable-like C-V-C unit) for study and further subdivided the unit into portions to be considered as potential phonemes. The results of this investigation were tested and additional rules developed using a limited vocabulary word list. Time compression permits real-time spectrum analysis with a sweep analyzer and offers the possibility of a reduction in equipment size approaching an order of magnitude compared with conventional parallel-channel analyzers. Analyses were carried out to show the theoretical capabilities of time compression and general criteria were organized for design under various performance requirements. Feasibility models of two methods of time compression, one using a solid-state analog memory and the other using a novel magnetic drum, were constructed and studied. Results indicate that both methods are basically sound. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 30, 1961
- Accession Number
- AD0272109
Entities
People
- R.w. Freytag
- W.d. Larkin
Organizations
- General Dynamics