Transcribing Lectures and Seminars ∗
Transcribing Lectures and Seminars ∗
复制标题
抄写讲座和研讨会 *
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
J. Gauvain
中科院分区：
文献类型：
--
作者：
L. Lamel;G. Adda;Éric Bilinski;J. Gauvain
本文介绍了在 FP6 集成项目 C HIL 的背景下进行的最新研究，旨在开发自动转录讲座和研讨会的系统。我们利用广泛可用的语料库来训练声学和语言模型，因为只有少量的 CHIL 数据可用于系统开发。对于声学模型训练，使用了 Eurospeech 录音的 TED 语料库的转录部分，以及 ICSI、ISL 和 NIST 会议语料库。对于语言模型训练，文本材料是从各种在线会议记录中提取的。根据 12 个研讨会提取的测试数据，得出约 25% 的单词错误率。
This paper describes recent research carried out in the context of the FP6 Integrated Project C HIL in developing a system to automatically transcribe lectures and seminars. We made use of widely available corpora to train both the acoustic and language models, since only a small amount of C HIL data were available for system development. For acoustic model training made use of the transcribed portion of the TED corpus of Eurospeech recordings, as well as the ICSI, ISL, and NIST meeting corpora. For language model training, text materials were extracted from a variety of on-line conference proceedings. Word error rates of about 25% are obtained on test data extracted 12 seminars.
影响因子：
4.3
作者：
LEGGETTER, CJ;WOODLAND, PC
通讯作者：
WOODLAND, PC