Automatic Spoken Language Identification Using Emotional Speech

Spoken language identification (LID) is the process of automatically recognizing the language from the uttered speech of an unknown speaker. Automatic recognition of language spoken is of vital importance in human-computer interaction and its applications. It can be applied in speech-to-speech translation systems, at call centers to re-route incoming calls to native speaker operators, and in speaker diarization in multilingual environments. The majority of studies which utilized LID systems focused solely on the use of neutral (i.e., normal) speech. However, in real applications and for comprehensive research investigations, the use of emotional speech in LID is crucial. The current study aims at investigating the effectiveness and performance of a deep neural networks (DNNs) based LID system when emotional speech is used.