A trajectory mixture density network for the acoustic-articulatory inversion mapping
A trajectory mixture density network for the acoustic-articulatory inversion mapping
复制标题
DOI：
10.21437/interspeech.2006-213
复制
发表时间：
2006-09
期刊：
影响因子：
--
通讯作者：
Korin Richmond
中科院分区：
文献类型：
--
作者：
Korin Richmond
本文提出了一种轨迹模型，它是基于一个混合密度网络训练的目标特征增强动态功能与算法估计最大似然轨迹尊重静态和派生的动态功能之间的约束。该模型进行了评价的反演映射任务。我们发现，轨迹模型的引入成功地将均方根误差降低了7.5%，并提高了相关性得分。索引术语：声学发音反演，条件轨迹模型，混合密度网络。
This paper proposes a trajectory model which is based on a mixture density network trained with target features augmented with dynamic features together with an algorithm for estimating maximum likelihood trajectories which respects constraints between the static and derived dynamic features. This model was evaluated on an inversion mapping task. We found the introduction of the trajectory model successfully reduced root mean square error by up to 7.5%, as well as increasing correlation scores. Index Terms: acoustic-articulatory inversion, conditional trajectory model, mixture density network.