arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2204.03315cs.CLcs.SDeess.AS
收藏

基于预训练DNN-HMM声学-语音模型的端到端口语语言理解三模块建模

Three-Module Modeling For End-to-End Spoken Language Understanding Using Pre-trained DNN-HMM-Based Acoustic-Phonetic Model

Ping An Technology(平安科技);Yanshan University(燕山大学)

Nick J. C. Wang, Lu Wang, Yandan Sun, Haimei Kang, Dejun Zhang

摘要:

在口语语言理解(SLU)中,用户所说的话被转换为他/她的意图。最近关于端到端SLU的工作表明,可以通过预训练方法提高准确性。我们使用语音预训练和三模块建模重新审视Lugosch等人提出的想法;但是,为了简化端到端SLU模型的构建,我们使用来自DNN-HMM的开源声学语音模型作为音素模块混合自动语音识别(ASR)系统,而不是从头开始训练。因此,我们只对单词模块的语音进行微调,并在单词和意图模块上应用多目标学习(MTL)来共同优化SLU性能。MTL在意图分类出错率方面相对降低了40%(从1.0%降低到0.6%)。请注意,我们的三模块模型是一个流方法。所提出的三模块建模方法的最终结果在FluentSpeech上产生了99.4%的意图准确性,与Lugosch等人相比,意图出错率降低了50%。虽然我们专注于实时流方法,但我们也列出了非流方法进行比较。

英文摘要:

In spoken language understanding (SLU), what the user says is converted to his/her intent. Recent work on end-to-end SLU has shown that accuracy can be improved via pre-training approaches. We revisit ideas presented by Lugosch et al. using speech pre-training and three-module modeling; however, to ease construction of the end-to-end SLU model, we use as our phoneme module an open-source acoustic-phonetic model from a DNN-HMM hybrid automatic speech recognition (ASR) system instead of training one from scratch. Hence we fine-tune on speech only for the word module, and we apply multi-target learning (MTL) on the word and intent modules to jointly optimize SLU performance. MTL yields a relative reduction of 40% in intent-classification error rates (from 1.0% to 0.6%). Note that our three-module model is a streaming method. The final outcome of the proposed three-module modeling approach yields an intent accuracy of 99.4% on FluentSpeech, an intent error rate reduction of 50% compared to that of Lugosch et al. Although we focus on real-time streaming methods, we also list non-streaming methods for comparison.

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。

↑