arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2107.01569cs.CLcs.LG
收藏

基于交叉模态变换的语音识别神经校正模型

Cross-Modal Transformer-Based Neural Correction Models for Automatic Speech Recognition

NTT Media Intelligence Laboratories, NTT Corporation(日本电信电话株式会社媒体智能研究所)

Tomohiro Tanaka, Ryo Masumura, Mana Ihori, Akihiko Takashima, Takafumi Moriya, Takanori Ashihara, Shota Orihashi, Naoki Makishima

摘要:

提出了一种基于跨模态变换的神经校正模型,该模型对自动语音识别(ASR)系统的输出进行改进,以排除ASR错误。一般来说,神经校正模型由编码器-解码器网络组成,可以直接模拟序列到序列的映射问题。最成功的方法是使用输入语音及其ASR输出文本作为编码器-解码器网络的输入上下文。然而,传统的方法不能考虑这两个不同的模态输入之间的关系,因为输入上下文被单独编码为每个模态。为了有效地利用两个不同模态输入之间的相关信息,我们提出的模型编码两个不同的上下文联合使用Transformer的跨模态自我注意的基础上。我们期望跨通道自我注意能有效地捕捉两个不同通道之间的关系,以完善ASR假设。我们还引入了一种浅层融合技术,有效地整合了第一遍ASR模型和我们提出的神经校正模型。日语自然语言ASR任务的实验表明,我们提出的模型实现了更好的ASR性能比传统的神经校正模型。

英文摘要:

We propose a cross-modal transformer-based neural correction models that refines the output of an automatic speech recognition (ASR) system so as to exclude ASR errors. Generally, neural correction models are composed of encoder-decoder networks, which can directly model sequence-to-sequence mapping problems. The most successful method is to use both input speech and its ASR output text as the input contexts for the encoder-decoder networks. However, the conventional method cannot take into account the relationships between these two different modal inputs because the input contexts are separately encoded for each modal. To effectively leverage the correlated information between the two different modal inputs, our proposed models encode two different contexts jointly on the basis of cross-modal self-attention using a transformer. We expect that cross-modal self-attention can effectively capture the relationships between two different modals for refining ASR hypotheses. We also introduce a shallow fusion technique to efficiently integrate the first-pass ASR model and our proposed neural correction model. Experiments on Japanese natural language ASR tasks demonstrated that our proposed models achieve better ASR performance than conventional neural correction models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑