Automatic Spoken Language Identification
Maged Ahmed Maher Nofal;
Abstract
(language ID for short) is the problem of identifying the language being spoken from a sample of speech uttered by an unknown speaker.
The thesis describes the importance of having an automatic spoken language identification system. It discusses different approaches used to develop language identification systems for telephone quality, short duration speech utterances, both early and recent approaches are discussed. The thesis reviews basic issues in speech signal processing beginning from the concept ofdigital speech till obtaining a vector of features from any speech frame. The thesis also reviews the fundamentals of speech recognition based on Hidden Markov Models, and discusses how to train Hidden Markov Models and then use them in recognition.
The thesis proposes different designs and implementations of an Arabic/English language identification system based on speaker identification approach. It implements the main components used in building a language identification system, many variations of the
•design of these components are implemented and tested and the results of their testing are compared with each other. This includes testing the effect of vowel spotting on both speaker and language identification.
The thesis describes the importance of having an automatic spoken language identification system. It discusses different approaches used to develop language identification systems for telephone quality, short duration speech utterances, both early and recent approaches are discussed. The thesis reviews basic issues in speech signal processing beginning from the concept ofdigital speech till obtaining a vector of features from any speech frame. The thesis also reviews the fundamentals of speech recognition based on Hidden Markov Models, and discusses how to train Hidden Markov Models and then use them in recognition.
The thesis proposes different designs and implementations of an Arabic/English language identification system based on speaker identification approach. It implements the main components used in building a language identification system, many variations of the
•design of these components are implemented and tested and the results of their testing are compared with each other. This includes testing the effect of vowel spotting on both speaker and language identification.
Other data
| Title | Automatic Spoken Language Identification | Other Titles | التعرف الآلي على اللغة المنطوقة | Authors | Maged Ahmed Maher Nofal | Issue Date | 1999 |
Recommend this item
Similar Items from Core Recommender Database
Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.