爱哭鬼

1.root权限下输入指令：sudo apt install ffmpeg2.检查版本ffmpeg -version显示以下信息则安装成功

1.源码16k采样率模型训练参考Github:https://github.com/YongyuG/rnnoise_16k2.编译特征提取工具编译前，进入denoise.c, 修改count参数，数值根据你的语音文件大小选定，第一次可以选小一点，体验流程if (count==1000000) break；cd rnnoise-master/src./compile.sh执行后会生成文件denoise_training，使用方式如下：../src/denoise_training_gao

语音增强:https://shichaog1.gitbooks.io/hand-book-of-speech-enhancement-and-recognition/content/chapter13.html

# 查看keras版本import h5pymodelPath = "./weights.hdf5"f = h5py.File(modelPath, 'r')print(f.attrs.get('keras_version'))

1.源码Github:https://github.com/xiph/rnnoise2.