asr_thchs30 使用THCHS30 中文语音数据集训练了语言模型,并使用该模型进行语音识别。 模型文件 下载地址: 链接:https://pan.baidu.com/s/1y-tEIA34e48t38uSNzV33g 提取码:asrm 下载模型文件后,放在同一文件夹即可。 loss曲线: librosa版本问题 这里使用的 librosa==0.7.2 可能会出现 参考这份博客: 解决不联网环境pip安装librosa、numba、llvmlite报错和版本兼容问题 修改如下: