Language identification with language and feature dependency

The purpose of Language Identification (LID) is to identify a specific language from aspoken utterance, automatically. Language-specific characteristics are always associatedwith different languages. Most existing LID approaches utilise a statistical modellingprocess with common acoustic/phonotactic features to model specific languages whileavoiding any language-specific knowledge. Great successes have been achieved in thisarea over past decades. However, there is still a huge gap between these languageindependentmethods and the actual language-specific patterns. It is extremely useful toaddress these specific acoustic or semantic construction patterns, without spending hugelabour on annotation which requires language-specific knowledge. Inspired by this goal,this research focuses on the language-feature dependency. Several practical methodshave been proposed.Various features and modelling techniques have been studied in this research. Someof them carry out additional language-specific information without manual labelling,such as a novel duration modelling method based on articulatory features, and a novelFrequency-Modulation (FM) based feature. The performance of each individual featureis studied for each of the language-pair combinations. The similarity between languagesand the contribution in identifying a language by using a particular feature are definedfor the first time, in a quantitative style. These distance measures and languagedependentcontributions become the foundations of the later-presented frameworks –language-dependent weighting and hierarchical language identification. The latterparticularly provides remarkable flexibility and enhancement when identifying arelatively large number of languages and accents, due to the fact that the mostdiscriminative feature or feature-combination is used when separating each of thelanguages.The proposed systems are evaluated in various corpora and task contexts includingNIST language recognition evaluation tasks. The performances have been improved invarious degrees.The key techniques developed for this work have also been applied to solve adifferent problem other than LID – speech-based cognitive load monitoring