Paper proposes i-vector approach for NLI with improved accuracy.
problem Identifying a speaker's native language from speech in a second language.
method i-vector approach using MFCC and GFCC features.
result Improved accuracy in NLI system, especially with GFCC features.