Text independent gender identification in noisy environmental conditions

摘要

This paper proposes a competent system that is not only text independent in identifying gender of a speaker but can also work efficiently in noisy environmental conditions in real time. The noisy environmental conditions are the places where noise signals are generated at different SNRs (Signal to Noise Ratios) such as train station, restaurant, exhibition hall, airport, and so on. The algorithms used in the proposed system are MFCC (Mel-Frequency Cepstral Coefficients) for feature extraction from the speech and ANN (Artificial Neural Network) for classification between the genders (Male and Female).