ParallelProcessor usage

[keithchan9999]

Expected behaviour

Please describe how you expected madmom to behave.

Actual behaviour

I am trying to run both a beat processor and a tempo processor which wraps under a parallel processor. But I got an error like this:

File "mad.py", line 158, in main
madmom.processors.process_online(parallel_processor, **kwargs)
File "C:\Users\kido9\miniconda3\envs\mad\lib\site-packages\madmom\processors.py", line 885, in process_online
_process((processor, frame, outfile, process_args))
File "C:\Users\kido9\miniconda3\envs\mad\lib\site-packages\madmom\processors.py", line 289, in _process
return process_tuple[0](*process_tuple[1:-1], **process_tuple[-1])
File "C:\Users\kido9\miniconda3\envs\mad\lib\site-packages\madmom\processors.py", line 120, in call
return self.process(*args, **kwargs)
TypeError: process() takes 2 positional arguments but 3 were given

Tried to troubleshoot but unfortunately I ain't any expert on that.

Steps needed to reproduce the behaviour

kwargs = dict(
		fps = 100,
		correct = True,
		infile = None,
		outfile = None,
		max_bpm = 150,
		min_bpm = 60,
		num_frames = 1,
		online = True,
		stream_input_device = dev_index,
		#nn_files = [BEATS_LSTM[0], BEATS_LSTM[1], BEATS_LSTM[2], BEATS_LSTM[3], BEATS_LSTM[4]]
		#verbose = 1
	)

	in_processor = madmom.features.beats.RNNBeatProcessor(**kwargs)
	beat_processor = madmom.features.beats.DBNBeatTrackingProcessor(**kwargs)
	out_processor = [beat_processor, beat_callback]
	processor = madmom.processors.IOProcessor(in_processor, out_processor)

	tempo_processor = madmom.features.tempo.TempoEstimationProcessor(**kwargs)
	out2_processor = [tempo_processor, tempo_callback]
	processor2 = madmom.processors.IOProcessor(in_processor, out2_processor)

	parallel_processor = madmom.processors.ParallelProcessor([processor, processor2], num_threads=2)
	madmom.processors.process_online(parallel_processor, **kwargs)

Information about installed software

Please provide some information about installed software.

import madmom
import numpy as np
import scipy

madmom.__version__
np.__version__
scipy.__version__

[superbock]

madmom does not have such a sophisticated process management, so you need to do this by yourself, i.e. wrap the beat tempo sub-pipelines in a parallel processor and use in_processor to feed it.

in_processor = madmom.features.beats.RNNBeatProcessor(**kwargs)

beat_tracker = madmom.features.beats.DBNBeatTrackingProcessor(**kwargs)
beat_processor = madmom.processors.IOProcessor(beat_tracker, beat_callback)

tempo_estimator = madmom.features.tempo.TempoEstimationProcessor(**kwargs)
tempo_processor = madmom.processors.IOProcessor(tempo_estimator, tempo_callback)

parallel_processor = madmom.processors.ParallelProcessor([beat_processor, tempo_processor], num_threads=2)

pipeline = madmom.processors.SequentialProcessor(in_processor, parallel_processor)
madmom.processors.process_online(pipeline, **kwargs)

However, you have to make sure the your beat_callback and tempo_callback accept two arguments (data, output).

HTH,