arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2003.07705eess.AScs.CLcs.LGcs.SD
收藏

混合自回归传感器(帽子)

Hybrid Autoregressive Transducer (hat)

Google(谷歌公司)

Ehsan Variani, David Rybach, Cyril Allauzen, Michael Riley

摘要:

本文提出并评估了混合自回归换能器(HAT)模型,时间同步encoderdecoder模型,保留了传统的自动语音识别系统的模块化。HAT模型提供了一种衡量内部语言模型质量的方法,可以用来决定使用外部语言模型进行推理是否有益。本文还提出了HAT模型的有限上下文版本,它解决了暴露偏差问题,并显着简化了整体训练和推理。我们评估我们提出的模型上的一个大规模的语音搜索任务。我们的实验表明,WER显着改善相比,国家的最先进的方法。

英文摘要:

This paper proposes and evaluates the hybrid autoregressive transducer (HAT) model, a time-synchronous encoderdecoder model that preserves the modularity of conventional automatic speech recognition systems. The HAT model provides a way to measure the quality of the internal language model that can be used to decide whether inference with an external language model is beneficial or not. This article also presents a finite context version of the HAT model that addresses the exposure bias problem and significantly simplifies the overall training and inference. We evaluate our proposed model on a large-scale voice search task. Our experiments show significant improvements in WER compared to the state-of-the-art approaches.

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。

↑