[hts-users:00216] Re: about prosodic model

[Keiichi Tokuda <tokuda@xxxxxxxxxxxxxxxx>]

From: 刘 磊 <liulei_198216@xxxxxxxxxxx>
Subject: [hts-users:00215] about prosodic model
Date: Wed, 15 Mar 2006 14:51:19 +0800
Message-ID: <BAY7-F9711006BE2D9CE33D08FE85E60@xxxxxxx>

> There is a viewpoint that 
> HTS handles
> F0's on a frame by frame basis, it may cause sudden F0 undulations,
> which are not characteristic of human
> speech. Prosodic features cover a wider time span
> than segmental features, and should be treated
> differently.

That is a reason why we are using the relationship between
static and dynamic feature sequences in the speech parameter
generation from HMM.

Keiichi Tokuda
tokuda@xxxxxxxxxxxxxxxx
http://kt-lab.ics.nitech.ac.jp/~tokuda/