arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2004.11544eess.AS
收藏

实现快速准确的端到端流传输ASR

Towards Fast and Accurate Streaming End-to-End ASR

Bo Li, Shuo-yiin Chang, Tara N. Sainath, Ruoming Pang, Yanzhang He, Trevor Strohman, Yonghui Wu

摘要:

端到端(E2 E)模型将传统语音识别模型的声学、发音和语言模型折叠成一个神经网络,其参数数量比传统ASR系统少得多,从而使其适合于设备上的应用。例如,作为流E2 E模型的递归神经网络换能器(RNN-T)已经显示出用于设备上ASR的有希望的潜力。对于这样的应用,质量和延迟是两个关键因素。我们建议通过扩展RNN-T端点(RNN-T EP)模型来减少E2 E模型的延迟,并增加早期和晚期惩罚。通过进一步应用最小字错误率(MWER)训练技术,我们在语音搜索测试集上实现了8.0%的相对字错误率(WER)降低和130 ms的90百分位延迟降低。我们还尝试了第二遍Listen,Attend and Spell(LAS)重新评分器。虽然它没有直接改善第一遍延迟,但大的WER降低提供了额外的空间来用WER换取延迟。与最初提出的RNN-T EP模型相比,RNN-T EP+LAS加上MWER训练可使相对WER减少18.7%,并使90百分位数延迟减少160 ms。

英文摘要:

End-to-end (E2E) models fold the acoustic, pronunciation and language models of a conventional speech recognition model into one neural network with a much smaller number of parameters than a conventional ASR system, thus making it suitable for on-device applications. For example, recurrent neural network transducer (RNN-T) as a streaming E2E model has shown promising potential for on-device ASR. For such applications, quality and latency are two critical factors. We propose to reduce E2E model's latency by extending the RNN-T endpointer (RNN-T EP) model with additional early and late penalties. By further applying the minimum word error rate (MWER) training technique, we achieved 8.0% relative word error rate (WER) reduction and 130ms 90-percentile latency reduction over on a Voice Search test set. We also experimented with a second-pass Listen, Attend and Spell (LAS) rescorer . Although it did not directly improve the first pass latency, the large WER reduction provides extra room to trade WER for latency. RNN-T EP+LAS, together with MWER training brings in 18.7% relative WER reduction and 160ms 90-percentile latency reductions compared to the original proposed RNN-T EP model.

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。

↑