[hts-users:03269] Re: Shared Decision Tree Context Clustering of Speaker

[hts-users:03269] Re: Shared Decision Tree Context Clustering of Speaker Adaptive Training

Subject: [hts-users:03269] Re: Shared Decision Tree Context Clustering of Speaker Adaptive Training

Date: Tue, 24 Apr 2012 02:30:26 +0800

Delivered-to: hts-users@xxxxxxxxxxxxxxx

Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type; bh=AK9MTFoMYD10aaP2AczEJcpzNNBnXYGm6CgZKFWQtw4=; b=wroOPzpxy+vRAoDQT5VvnwVIEZqQOXTCuSWBCe0rm0yvYdALLUxaHAhf4N2P99UNGx rP+jkNCheZf7MOAq8JXVIf8Xo0Adh7H4nJgCNpjGV6r1wMh2q8H+SZn+1Szywyd/n00M abul+5rFFzABAYUn/Kmm0fkfzq9jaCYLkijinxq6pxFNdKmoHg+OXH/QkQQSlJdoCbeG rbuamr7tqzOgBD8WzhRmyMnvKKEXMJYmNuxCao7MaULdojuehUhTsaT5ZQoIyo8mAIuR uNyHg+NZTqfJSlh2eC6zdUBEx8aYgijDarWvx5w1oVZeedOlMV9TBkpUOytuUz1z+JUM uOVA==

Hi,

Is there any way to use shared decision tree with HTS? As to average voiced modeling shared decision tree seems to be more efficient than speaker adaptive training(SAT) according to the paper of Dr. Yamagishi, and these two methods are the main methods to improve the performance of the average voice model.

Jay

那兴宇 <nxy-yzqs@xxxxxxx> 於 2012年4月21日下午10:15 ?道：

Hi,

Shared decision tree is not applied in current HTS packages.

--

Xingyu Na (那兴宇)

Beijing Institute of Technology

naxy(at)bit.edu.cn

asr.naxingyu(at)gmail.com

naxingyu at {facebook, twitter, linkedin}

At 2012-04-21 19:46:45,"li jay" <lij.acd@xxxxxxxxx> wrote:

Hi,

I'm studying speaker adaptive training and reading a paper of Dr. Yamagishi.

According to his paper the speaker adaptive training uses shared decision tree context clustering instead of normal decision tree based context clustering.

I traced the HTS-demo_CMU-ARCTIC-ADAPT demo Training.pl script and I found that when the demo system perform the speaker adaptive training, it doesn't use shared decision tree context clustering. Instead, it uses normal decision tree based context clustering. Is there any option so that HTS can use shared decision tree context clustering when performing speaker adaptive training?

Jay

网易Lofter，专注兴趣，分享创作！