Kannada Language Parameters for Speaker Identification with The Constraint of Limited Data
Citations Over TimeTop 22% of 2013 papers
Abstract
In this paper we demonstrate the impact of language parameter variability on mono, cross and mult ilingual speaker identificat ion under limited data condition. The languages considered for the study are English, Hindi and Kannada. The speaker specific features are ext racted using multi-taper mel-frequency cepstral coefficients (MFCC) and speaker models are built using Gaussian mixture model (GMM)-universal background model (UBM). The sine-weighted cepstrum estimators (SWCE) with 6 tapers are considered for mu lti-taper MFCC feature ext raction. The mono and cross-lingual experimental results show that the performance of speaker identificat ion trained and/or tested with Kannada language is decreased as co mpared to other languages. It was observed that a database free fro m ottakshara, arka and anukaranavyayagalu results a good performance and is almost equal to other languages.
Related Papers
- → Infant cry recognition based on feature extraction(2010)3 cited
- → Infant asphyxia detection using autoencoders trained on locally linear embedded-reduced Mel Frequency Cepstrum Coefficient (MFCC) features(2018)2 cited
- → Pitch-based cepstral features for gender classification in noisy environments(2013)1 cited
- A Robust Mel-frequency Cepstrum Coefficients(2008)
- Application of Biomimetic Technology to Feature Extraction from Acoustic Objects(2014)