TinyML之micro_speech语音识别----Feature generation failed解决方案

具体的开发过程网上都有，就不赘述了。

在colab上训练模型，花了将近2小时，得到模型文件后，CV到micro_speech工程里的micro_features_model.cpp里，结果运行报错：

Feature generation failed
Requested feature_data_ size 536907080 doesn't match 1960

最终在github上找到了解决方法，记录一下，在micro_speech.ino中作如下修改：

  static tflite::MicroMutableOpResolver<5> micro_op_resolver(error_reporter);
  if (micro_op_resolver.AddDepthwiseConv2D() != kTfLiteOk) {
    return;
  }
  if (micro_op_resolver.AddFullyConnected() != kTfLiteOk) {
    return;
  }
  if (micro_op_resolver.AddSoftmax() != kTfLiteOk) {
    return;
  }
  if (micro_op_resolver.AddReshape() != kTfLiteOk) {
    return;
  }
  if (micro_op_resolver.AddConv2D() != kTfLiteOk) { 
    return; 
  }

其实就是增加了

if (micro_op_resolver.AddConv2D() != kTfLiteOk) { return; }

并把

static tflite::MicroMutableOpResolver<4> micro_op_resolver(error_reporter);

改成

static tflite::MicroMutableOpResolver<5> micro_op_resolver(error_reporter);