Transducer-based language embedding for spoken language identification
Transducer-based language embedding for spoken language identification
复制标题
DOI：
10.48550/arxiv.2204.03888
复制
发表时间：
2022-04
期刊：
影响因子：
--
通讯作者：
Peng Shen;Xugang Lu;H. Kawai
中科院分区：
文献类型：
--
作者：
Peng Shen;Xugang Lu;H. Kawai
声学和语言特征是口语识别（LID）任务的重要线索。最近先进的 LID 系统主要使用声学特征，缺乏显式语言特征编码的使用。在本文中，我们通过将 RNN 转换器模型集成到语言嵌入框架中，提出了一种用于 LID 任务的新型基于转换器的语言嵌入方法。受益于 RNN 传感器的语言表示能力的优势，所提出的方法可以利用语音感知的声学特征和显式语言特征来完成 LID 任务。在大规模多语言 LibriSpeech 和 VoxLingua107 数据集上进行了实验。实验结果表明，该方法显着提高了 LID 任务的性能，在域内和跨域数据集上分别相对提高了 12% 到 59% 和 16% 到 24%。
The acoustic and linguistic features are important cues for the spoken language identification (LID) task. Recent advanced LID systems mainly use acoustic features that lack the usage of explicit linguistic feature encoding. In this paper, we propose a novel transducer-based language embedding approach for LID tasks by integrating an RNN transducer model into a language embedding framework. Benefiting from the advantages of the RNN transducer's linguistic representation capability, the proposed method can exploit both phonetically-aware acoustic features and explicit linguistic features for LID tasks. Experiments were carried out on the large-scale multilingual LibriSpeech and VoxLingua107 datasets. Experimental results showed the proposed method significantly improves the performance on LID tasks with 12% to 59% and 16% to 24% relative improvement on in-domain and cross-domain datasets, respectively.