English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

— In the present study, three language identification (LID) approaches are compared for a set of three Indian languages. However, the number of languages can be increased. One of the approaches is the Hidden Markov Model (HMM) based Phonetic Engine (PE), another is the Gaussian Mixture Model based Universal Background Model (GMM-UBM) classifier and the remaining one is a prosodic feature based classification system. The PE belongs to the category of Explicit LID techniques while the GMM-UBM system and the prosodic feature based classifier fall into the group of Implicit LID techniques. Basically, Explicit LID techniques require a segmented and phonetically labelled speech corpus, while the Implicit LID techniques do not require any phonetic description of the data. All the systems are used here in identifying a set of test data from three Indian languages, Manipuri, Assamese and Bengali. The selection of these languages is made on the basis of the wide range of usages of them in North Eastern India and at the same time no identification tasks were carried out so far for a database inclusive of these three languages together. The purpose of this comparison is to check the LID efficiency among the three approaches. In the experiments, it is found that the prosodic feature based approach can only be used for a broader classification of languages into two categories such as tonal and non tonal. Such a system can’t be used alone to completely identify a language. However, the other two approaches can suitably be used for automatic identification of the languages. The identification rate (IDR) of the PE is superior to the GMMUBM classifier. The average IDR reported with PE is 99% while for the GMM-UBM classifier it is found to be 96.94% with the same speech corpus being in use. However, the data preparation task is a little more expensive in PE than that of GMM-UBM classifier. Thus, the compensation for accuracy may be paid with the cost incurred in data preparation.

A Comparative evaluation of three automatic language identification approaches for Indian Languages