Graduate Thesis Or Dissertation
 

Pattern recognition of speech using hybrid computer feature extraction

Öffentlich Deposited

Herunterladbarer Inhalt

PDF Herunterladen
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/0z709109w

Descriptions

Attribute NameValues
Creator
Abstract
  • Most technological problems associated with man-machine conversations have been solved and are well documented in both the contemporary technical and lay literature. The major remaining technological problem is the real time conversion of a human's vocal uterances into a "written" phoneme sequence representing the information content of speech. The work presented here demonstrates a viable solution of the problem of real time machine determination of the information content of human speech. The solution presented differs extensively from previous attempts to solve this problem which used relatively crude template matching techniques. The techniques reported here are not limited to a fixed vocabulary or a master's voice as previous attempts have been. They are self-adapting to each individual's speech patterns regardless of dialect, nationality, or language. This system is capable of determining the information content present in human uterances with a greater degree of accuracy than a human listener can perceive speech, disregarding contextual information. Also, human voice perception is not limited by noise masking to the extent exhibited by the human ear. Operationally, the real time, hybrid computer system first estimates the model parameters of the human speech generation mechanism from phonetic features extracted from the speech waveform. The system next is able to deterministically relate the estimates of the human's speech generation model parameters to the phonemes (the information content) present in the human's vocal utterance. The speech recognition process is a multi-level process directly analogous to the multi-level speech recognition method used by a human listener.
Resource Type
Date Available
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Academic Affiliation
Non-Academic Affiliation
Subject
Urheberrechts-Erklärung
Publisher
Peer Reviewed
Language
Digitization Specifications
  • File scanned at 300 ppi (Monochrome) using ScandAll PRO 1.8.1 on a Fi-6770A in PDF format. CVista PdfCompressor 5.0 was used for pdf compression and textual OCR.
Replaces

Beziehungen

Parents:

This work has no parents.

In Collection:

Artikel

Miniaturansicht Titel Hochladedatum Sichtbarkeit Aktionen
file details HerrAlanJ1975.pdf 2017-08-23 Öffentlich Herunterladen