针对发音错误检测的发音字典生成提出基于联合序列多阶模型（Joint-sequence multi-gram,JSM）和多层神经感知（Multi-layer perception,......