What does the output of onsets_frames_wavinput.tflite mean?

Hello,

I am developing a piano education application using a flutter.

I'm a front-end developer and my knowledge of machine learning is not very deep.

Currently in my App, Can successfully load model and return outputs of onsets_frames_wavinput.tflite on Android/iOS.

Here some code (android)

      Object[] inputArray = {floatInputBuffer}; // size : 17920
      Map<Integer, Object> outputMap = new HashMap<>();
      float[][][] floatOutputBuffer = new float[1][32][88];
      outputMap.put(0, floatOutputBuffer);

      tfLite.runForMultipleInputsOutputs(inputArray,outputMap);  // The outputMap is an array of size 1x32x88.

88 seems to mean the number of piano keys. But I don't know what the 32 in the array means. Does 32 mean that the input buffer was processed for 32ms?

A lot of people seem to be wondering how to utilize this model in mobile applications.

I want to create an open source that successfully processes output and share it with people.