[Python] how to convert a python np.ndarray to Java Float[] array [duplicate]

I am using py4j for a project to connect my Python code to a JVM. Now this project requires me to pass an audio signal from the Python end to the Java end. I am using Librosa to create the audio signal array from the audio file. This gives me a float64 numpy array. Now I want to pass this audio signal array as a parameter of a Java function using Py4J. Now this seems to be the problematic part due to difference in data types I assume.

I tried to pass it after converting it to a byte array, but although the dimension of the array remains the same the values somehow all changed making the results of my projects very inaccurate.

Python side:

audio, sr = librosa.load(r"Audio/snoring.wav", sr=16000)
byte_array = audio.tobytes()
mel2 = gateway.entry_point.checkSignal()


Java side:

public void checkSignal(byte[] byteArray){
FloatBuffer floatBuffer = ByteBuffer.wrap(byteArray).asFloatBuffer();
float[] floatArray = new float[floatBuffer.remaining()];
floatBuffer.get(floatArray);

if(floatArray.length>0){
System.out.println("the signal has been received." + floatArray.length);
}

signal = floatArray;
}


Before that I had my audio file stored in the Java project directory and I would pass the audio file name from the Python side to the Java function, and as expected it worked flawlessly. But the need to pass the signal as a parameter is making this quite hard.

Continue reading...