search

scelesticsiva / speaker_recognition_GMM_UBM

A speaker recognition system which uses GMM-UBM for use in an Android application which helps in monitoring patients suffering from Schizophrenia.
51 stars 17 forks source link

Can you provide some feature which can be used to train the ubm and gmm ? #1

Closed wuqiangch closed 6 years ago

wuqiangch commented 6 years ago

Can you provide some feature which can be used to train the ubm and gmm ? Thanks!

scelesticsiva commented 6 years ago

MFCC features are used for speaker recognition since it capture information similar to human auditory system, I have included a script for extracting MFCC from audio into a csv file(instructions to use it are in README)

PS: It requires ffmpeg installed in your system for it to work