Articulatory-Acoustic-Feature-based Automatic Language Identification

Automatic language identification is one of the important topics in multilingual speech technology. Ideal language identification systems should be able to classify the language of speech utterances within a specific time before further processing by language-dependent speech recognition systems or monolingual listeners begins. Currently the best language identification systems are based on HMM-based speech recognition systems. However, with the cost of this low percentage error, comes an increase in computational complexity. This paper proposes an alternative way of using HMM-based speech recognition systems. Instead of using phoneme level acoustic models and n-gram language models, articulatory feature level acoustic models and n-gram language models are introduced. With this approach, the computational complexities of language identification systems are considerably reduced due to the fact that the size of the articulatory feature inventory is naturally smaller than that of the of phoneme inventory.