基于双向改进门控循环单元维吾尔语语音识别

Uyghur Speech Recognition Based on Bidirectional Improved Gated Recurrent Unit

为了能够提升语音识别的准确率,并有效降低训练模型的复杂度,提出了一种双向改进门控循环单元的声学模型语音识别方法。模型上移除重置门,在状态更新过程中采用ReLU激活函数并与前馈连接采用的BN算法有效结合,改进的模型可以降低模型的计算复杂度,加快模型收敛;采用双向的结构不仅可以有效帮助模型捕捉到过去和未来的语义时序信息,而且可以有效提升识别准确率。在THUYG-20维吾尔语数据集上实验结果表明,与基线传统深度神经网络进行对比,基于双向改进门控循环单元网络词错误率下降2.34%;与标准双向长短期记忆网络(LSTM)比较每个迭代周期平均训练时间减少13.4%。

In order to improve the accuracy of speech recognition and effectively reduce the complexity of training model,further in-depth research is needed in the model optimization.This paper presents an acoustic model based on bidirectional improved gated recurrent unit.The improved model can reduce the complexity of the model and accelerate the rapid convergence of the model.The bidirectional structure can not only effectively help the model capture the past and future time series information,but also effectively improve the recognition accuracy.The experimental results on thuyg-20 Uyghur corpus show that compared with the baseline traditional depth neural network,the absolute word error rate is reduced by 2.34%using bidirectional improved gated recurrent unit;the model can also reduces the per-epoch training time by 13.4%over standardb idirectional long short-term memory(LSTM)model.