Spoken Language Identification (LiD) Using Machine Learning

Spoken Language Identification is the process of detecting the language of an utterance by an anonymous speaker, irrespective of gender, accent and pronunciations. Implementation of an acoustic model for Spoken Language Identification is to be carried out in this project. The major task is to identify those features or parameters which could be used to clearly distinguish between languages. This acoustic model makes use of mean values of Mel Frequency Cepstral Coefficients (MFCC). The system uses Support Vector Machine (SVM) to the handle the problem of multi class classification.The project aims at detecting English, Japanese, French, Hindi, and Kannada.
This work was submitted as my final year bachelors degree project. The project also won the first place at the National Level Project Competition & Exhibition held at M S Ramaiah Institute of Technology in 2012.
The detailed report of this project is available here