#Update:
Kaldi now has SRE10 example., so please use the recipes in the latest kaldi. Thanks. If you need infomation on Kaldi's data format you can still refer to this document.
The subdirectories "v1" and so on are different iVector-based speaker
recognition recipes. The recipe in v1 demonstrates a standard approach
using a full-covariance GMM-UBM, iVectors, and a PLDA backend. The example
in v2 replaces the GMM of the v1 recipe with a time-delay deep neural
network.
Kaldi SRE2010
This baseline build on well-established iVector/PLDA speaker verification framework for SRE 2010 female tasks. This work is based on Kaldi SRE 2008 example. The result is reported as the clean iVector/PLDA baseline in [1] with average EER 3.50% over nine core conditions.
To run this experiment plese first check data, hardware and software requirements and then read this guide.
For information about kaldi data format and IO please read this document.
If you feel this baseline helpful kindly cite Kaldi as well as this paper:
[1] Steven Du, Xiong Xiao, Eng Siong Chng, "DNN FEATURE COMPENSATION FOR NOISE ROBUST SPEAKER VERIFICATION", ChinaSIP 2015