基本信息

---

下载

   友情提示：（无需注册或充值，赞助后即可获取资源下载链接）

---

     嘿，亲！知识可是无价之宝呢，但咱这精心整理的资料也耗费了不少心血呀。小小地破费一下，绝对物超所值哦！如有下载和支付问题，请联系我们QQ(微信同号)：813200300

本次赞助数额为： 2 元　

支付宝支付

微信支付

   源码介绍

---

声纹成分分析

This is a demo on how to use the Identity Toolbox for i-vector based speaker
recognition. A small scale task has been designed using artificially
generated features for 20 speakers. Each speaker has 10 sessions
(channels) and each session is 1000 frames long (10 seconds assuming 10 ms
frame increments).

There are 5 steps involved:
 
 1. training a UBM from background data
 2. learning a total variability subspace from background statistics
 3. training a Gaussian PLDA model with development i-vectors
 4. scoring verification trials with model and test i-vectors
 5. computing the performance measures (e.g., EER and confusion matrix)

微软MSR包下载