arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1705.05992cs.CL
收藏

循环神经网络声学模型的帧叠加与帧保持

Frame Stacking and Retaining for Recurrent Neural Network Acoustic Model

Alibaba Shenma Search(阿里巴巴神马搜索)

Xu Tian, Jun Zhang, Zejun Ma, Yi He, Juan Wei

摘要:

帧堆叠被广泛应用于端到端的神经网络训练,如连接主义时间分类(CTC),它导致更准确的模型和更快的解码。然而,如果解码器不变,它不适合于基于上下文相关状态声学模型的传统神经网络。在本文中,我们提出了一种新的帧保持方法,并应用于解码。该系统采用帧保留和帧堆叠相结合的方法,可以减少训练和解码的时间消耗。使用它的长短期记忆(LSTM)递归神经网络(RNN)实现了几乎线性的训练加速比,并降低了相对41%的实时的因子(RTF)。同时,在神马中文语音搜索数据集上,识别性能没有下降或略有提高。

英文摘要:

Frame stacking is broadly applied in end-to-end neural network training like connectionist temporal classification (CTC), and it leads to more accurate models and faster decoding. However, it is not well-suited to conventional neural network based on context-dependent state acoustic model, if the decoder is unchanged. In this paper, we propose a novel frame retaining method which is applied in decoding. The system which combined frame retaining with frame stacking could reduces the time consumption of both training and decoding. Long short-term memory (LSTM) recurrent neural networks (RNNs) using it achieve almost linear training speedup and reduces relative 41\% real time factor (RTF). At the same time, recognition performance is no degradation or improves sightly on Shenma voice search dataset in Mandarin.

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。

↑