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 1706.02776cs.CLcs.LGcs.NEstat.ML
收藏

语音识别中通过采样优化期望出错率

Optimizing expected word error rate via sampling for speech recognition

Google Inc.(谷歌公司)

Matt Shannon

摘要:

状态级最小贝叶斯风险(sMBR)训练已经成为语音识别声学模型序列级训练的事实标准。它有一个优雅的公式使用的期望半环,并给出了很大的改善字出错率(WER)的模型训练单独使用交叉熵(CE)或连接时间分类(CTC)。sMBR训练优化了参考声学状态和假设声学状态不同的预期帧数。优化期望WER可能是优选的,但是WER与期望半环不能很好地相互作用,并且基于计算期望WER的先前方法确切地涉及扩展在训练期间使用的格。在本文中,我们展示了如何通过从传统sMBR训练期间使用的网格中采样路径来优化预期的WER。预期WER的梯度本身是一个预期,因此可以使用蒙特卡洛采样来近似。我们的实验表明,在声学模型训练过程中优化WER,在2通道查询识别任务(Google Home)上,WER相对于调整良好的sMBR基线有5%的相对改善。

英文摘要:

State-level minimum Bayes risk (sMBR) training has become the de facto standard for sequence-level training of speech recognition acoustic models. It has an elegant formulation using the expectation semiring, and gives large improvements in word error rate (WER) over models trained solely using cross-entropy (CE) or connectionist temporal classification (CTC). sMBR training optimizes the expected number of frames at which the reference and hypothesized acoustic states differ. It may be preferable to optimize the expected WER, but WER does not interact well with the expectation semiring, and previous approaches based on computing expected WER exactly involve expanding the lattices used during training. In this paper we show how to perform optimization of the expected WER by sampling paths from the lattices used during conventional sMBR training. The gradient of the expected WER is itself an expectation, and so may be approximated using Monte Carlo sampling. We show experimentally that optimizing WER during acoustic model training gives 5% relative improvement in WER over a well-tuned sMBR baseline on a 2-channel query recognition task (Google Home).

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。

↑