Incremental Segmentation and Decoding Strategies for Simultaneous Translation
Incremental Segmentation and Decoding Strategies for Simultaneous Translation
复制标题
同声翻译增量分段和解码策略
DOI：
--
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
B. Sankaran
中科院分区：
文献类型：
--
作者：
M. Yarmohammadi;V. Sridhar;S. Bangalore;B. Sankaran
同声翻译是一项富有挑战性的任务，既要听源语言的语音，又要产生目标语言的语音。人类口译员通常毫不费力地完成这项任务，使用不同的策略，以最大限度地减少产生目标语言的延迟。为了模拟人类的翻译过程，我们提出了一种新的输入分割方法，使用短语对齐结构的语言对。我们比较和对比三种增量解码和两种不同的输入分割策略，包括我们提出的方法，同声翻译。我们提出的准确性和延迟权衡每个解码策略时，适用于音频讲座从TED集合。
Simultaneous translation is the challenging task of listening to source language speech, and at the same time, producing target language speech. Human interpreters achieve this task routinely and effortlessly, using different strategies in order to minimize the latency in producing target language. Toward modeling the human interpretation process, we propose a novel input segmentation method using the phrase alignment structure of the language pair. We compare and contrast three incremental decoding and two different input segmentation strategies, including our proposed method, for simultaneous translation. We present accuracy and latency tradeoffs for each of the decoding strategies when applied to audio lectures from the TED collection.