Changes for v0.17 release

.github/workflows/python-package-conda.yml

@@ -0,0 +1,73 @@
+name: Python Package using Conda
+
+on: [push]
+
+jobs:
+  build-linux:
+    runs-on: ubuntu-latest
+
+    defaults:
+      run:
+        shell: bash -l {0}
+
+    strategy:
+      max-parallel: 5
+      fail-fast: false
+      matrix:
+        python-version: [ "3.7", "3.8", "3.9", "3.10" ]
+
+    steps:
+    - uses: actions/checkout@v2
+
+    - name: Cache conda
+      uses: actions/cache@v2
+      env:
+        CACHE_NUMBER: 1  # increment this value to reset cache if environment.yml has not changed
+      with:
+        path: ~/conda_pkgs_dir
+        key: ${{ runner.os }}-${{ matrix.python-version }}-conda-${{ env.CACHE_NUMBER }}
+
+    - name: Install Conda env
+      uses: conda-incubator/setup-miniconda@v2
+      with:
+        auto-update-conda: true
+        python-version: ${{ matrix.python-version }}
+        add-pip-as-python-dependency: true
+        auto-activate-base: false
+        activate-environment: madmom
+        environment-file: environment.yml
+        use-only-tar-bz2: true  # this needs to be set for caching to work properly
+
+    - name: Conda info
+      run: |
+        conda info -a
+        conda list
+
+    - name: Install madmom
+      run: |
+        pip install -e .
+        git submodule update --init --remote
+
+    - name: Lint with flake8
+      run: |
+        # stop the build if there are Python syntax errors or undefined names
+        flake8 . --count --exit-zero --select=E9,F63,F7,F82 --show-source --statistics
+        # exit-zero treats all errors as warnings. The GitHub editor is 127 chars wide
+        flake8 . --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
+
+    - name: Test with pytest
+      run: |
+        pytest --cov --doctest-ignore-import-errors madmom tests
+
+    - name: Upload coverage to Codecov
+      uses: codecov/codecov-action@v1
+      with:
+        token: ${{ secrets.CODECOV_TOKEN }}
+        files: ./coverage.xml
+        directory: ./coverage/reports/
+        flags: unittests
+        env_vars: OS,PYTHON
+        name: codecov-umbrella
+        fail_ci_if_error: true
+        path_to_write_report: ./coverage/codecov_report.txt
+        verbose: true

README.rst

@@ -245,11 +245,15 @@ will give different help messages.
 Additional resources
 ====================
 
-Mailing list
-------------
+Discussions / Q&A
+-----------------
+
+Github discussions (https://github.com/CPJKU/madmom/discussions) should be
+used to get in touch with the developers and other users, to ask questions
+about madmom's usage and ideas for development.
 
-The `mailing list <https://groups.google.com/d/forum/madmom-users>`_ should be
-used to get in touch with the developers and other users.
+The `mailing list <https://groups.google.com/d/forum/madmom-users>`_ is kept
+as an archive but should not be used for new questions/
 
 Wiki
 ----
@@ -366,12 +370,17 @@ References
     *Deep Polyphonic ADSR Piano Note Transcription*,
     Proceedings of the 44th International Conference on Acoustics, Speech and
     Signal Processing (ICASSP), 2019.
+.. [20] Sebastian Böck, Matthew Davies and Peter Knees,
+    *Multi-Task learning of tempo and beat: learning one to improve the other*,
+    Proceedings of the 20th International Society for Music Information
+    Retrieval Conference (ISMIR), 2019.
+
 
 Acknowledgements
 ================
 
 Supported by the European Commission through the `GiantSteps project
-<http://www.giantsteps-project.eu>`_ (FP7 grant agreement no. 610591) and the
+<http://giantsteps-project.upf.edu>`_ (FP7 grant agreement no. 610591) and the
 `Phenicx project <http://phenicx.upf.edu>`_ (FP7 grant agreement no. 601166)
 as well as the `Austrian Science Fund (FWF) <https://www.fwf.ac.at>`_ project
 Z159.

bin/BarTracker

@@ -79,6 +79,7 @@ def main():
         # signal processor
         signal_processor = SignalProcessor(**vars(args))
         # beats processor
+        load_beats_processor = None
         if hasattr(args, 'infile'):
             # single mode: read the beats from STDIN
             load_beats_processor = LoadBeatsProcessor(**vars(args))

bin/evaluate

@@ -52,6 +52,11 @@ def main():
     # parse the args
     args = p.parse_args()
 
+    # print usage if no evaluation mode was set
+    if not getattr(args, 'eval', False):
+        p.print_usage()
+        exit(0)
+
     # print the arguments
     if args.verbose >= 2:
         print(args)

environment.yml

@@ -9,12 +9,17 @@ dependencies:
   - scipy
   - opencv
   - portaudio
+  - pyfftw
+  - ffmpeg
+  - pyaudio
   - pip
   - pip:
-      - mido>=1.2.6
-      - pyfftw
-      - pyaudio
-      - pytest
-      - black
-      - pre-commit
-      - prospector
+    - mido>=1.2.6
+    - flake8<4
+    - coverage
+    - pytest
+    - pytest-cov
+    - pytest-flake8
+    - black
+    - pre-commit
+    - prospector

madmom/audio/spectrogram.py

@@ -226,6 +226,9 @@ def tuning_frequency(self, **kwargs):
             Tuning frequency of the spectrogram.
 
         """
+        import warnings
+        warnings.warn('tuning_frequency() is deprecated as of version 0.17 '
+                      'and will be removed in 0.19.')
         from scipy.ndimage.filters import maximum_filter
         # widen the spectrogram in frequency dimension
         max_spec = maximum_filter(self, size=[1, 3])

madmom/evaluation/chords.py

@@ -54,7 +54,7 @@ def encode(chord_labels):
     Parameters
     ----------
     chord_labels : numpy structured array
-        Chord segments in `madmom.io.SEGMENT_DTYPE` format
+        Chord segments in `madmom.io.SEGMENT_DTYPE` format.
 
     Returns
     -------
@@ -728,10 +728,10 @@ class ChordEvaluation(EvaluationMixin):
 
     Parameters
     ----------
-    detections : str
-        File containing chords detections.
-    annotations : str
-        File containing chord annotations.
+    detections : numpy structured array
+        Detected chord segments in `madmom.io.SEGMENT_DTYPE` format.
+    annotations : numpy structured array
+        Annotated chord segments in `madmom.io.SEGMENT_DTYPE` format.
     name : str, optional
         Name of the evaluation object (e.g., the name of the song).
 

madmom/features/beats.py

@@ -1014,16 +1014,18 @@ def process_offline(self, activations, **kwargs):
             Detected beat positions [seconds].
 
         """
-        # init the beats to return and the offset
-        beats = np.empty(0, dtype=int)
+        # init beats to return and offset (to be added later)
+        beats = np.empty(0, dtype=float)
         first = 0
-        # use only the activations > threshold
+        # use only activations > threshold
         if self.threshold:
             activations, first = threshold_activations(activations,
                                                        self.threshold)
         # return no beats if no activations given / remain after thresholding
         if not activations.any():
             return beats
+        # add a small epsilon to prevent division by 0
+        activations += np.finfo(activations.dtype).eps
         # get the best state path by calling the viterbi algorithm
         path, _ = self.hmm.viterbi(activations)
         # also return no beats if no path was found

madmom/features/downbeats.py

@@ -121,8 +121,13 @@ def _process_dbn(process_tuple):
 
 class DBNDownBeatTrackingProcessor(Processor):
     """
-    Downbeat tracking with RNNs and a dynamic Bayesian network (DBN)
-    approximated by a Hidden Markov Model (HMM).
+    Downbeat tracking with a dynamic Bayesian network (DBN) approximated by
+    a Hidden Markov Model (HMM).
+
+    The observations must reflect the probabilities corresponding to beats
+    and downbeats. The probability for non-beats is computed given these two
+    probabilities as p(nb) = 1 - p(b) - p(db). All three observations must be
+    a probability density function, i.e. sum(p(nb) + p(b) + p(db)) == 1.
 
     Parameters
     ----------
@@ -268,21 +273,28 @@ def process(self, activations, **kwargs):
         """
         # pylint: disable=arguments-differ
         import itertools as it
-        # use only the activations > threshold (init offset to be added later)
+        # init beats to return and offset (to be added later)
+        beats = np.empty((0, 2), dtype=float)
         first = 0
+        # use only activations > threshold
         if self.threshold:
             activations, first = threshold_activations(activations,
                                                        self.threshold)
         # return no beats if no activations given / remain after thresholding
         if not activations.any():
-            return np.empty((0, 2))
+            return beats
+        # add a small epsilon to prevent division by 0
+        activations += np.finfo(activations.dtype).eps
         # (parallel) decoding of the activations with HMM
         results = list(self.map(_process_dbn, zip(self.hmms,
                                                   it.repeat(activations))))
         # choose the best HMM (highest log probability)
         best = np.argmax(list(r[1] for r in results))
         # the best path through the state space
         path, _ = results[best]
+        # also return no beats if no path was found
+        if not path.any():
+            return beats
         # the state space and observation model of the best HMM
         st = self.hmms[best].transition_model.state_space
         om = self.hmms[best].observation_model

madmom/ml/hmm.pyx

@@ -192,10 +192,8 @@ class TransitionModel(object):
         probabilities = np.asarray(probabilities)
         if not np.allclose(np.bincount(prev_states, weights=probabilities), 1):
             raise ValueError('Not a probability distribution.')
-        # convert everything into a sparse CSR matrix, make sure it is square.
-        # looking through prev_states is enough, because there *must* be a
-        # transition *from* every state
-        num_states = max(prev_states) + 1
+        # convert everything into a sparse CSR matrix, make sure it is square
+        num_states = max(states.max(), prev_states.max()) + 1
         transitions = csr_matrix((probabilities, (states, prev_states)),
                                  shape=(num_states, num_states))
         # convert to correct types

madmom/ml/nn/__init__.py

@@ -192,6 +192,9 @@ def load(cls, nn_files, **kwargs):
 
         """
         networks = [NeuralNetwork.load(f) for f in nn_files]
+        # raise error if no NNs were loaded
+        if not networks:
+            raise ValueError('No neural network(s) could be loaded.')
         return cls(networks, **kwargs)
 
     @staticmethod

madmom/processors.py

@@ -863,7 +863,7 @@ def process_online(processor, infile, outfile, **kwargs):
     kwargs['sample_rate'] = kwargs.get('sample_rate', 44100)
     kwargs['num_channels'] = kwargs.get('num_channels', 1)
     # list all available PyAudio devices and exit afterwards
-    if kwargs['list_stream_input_device']:
+    if kwargs.get('list_stream_input_device'):
         import pyaudio
         pa = pyaudio.PyAudio()
         for i in range(pa.get_device_count()):
@@ -947,6 +947,14 @@ def io_arguments(parser, output_suffix='.txt', pickle=True, online=False):
     # add general options
     parser.add_argument('-v', dest='verbose', action='count',
                         help='increase verbosity level')
+
+    # print usage if no processing mode is set
+    def print_usage(*args, **kwargs):
+        parser.print_usage()
+        exit(0)
+
+    parser.set_defaults(func=print_usage)
+
     # add subparsers
     sub_parsers = parser.add_subparsers(title='processing options')
 

setup.py

@@ -31,9 +31,15 @@
         ['madmom/features/beats_crf.pyx'],
         include_dirs=include_dirs,
     ),
-    Extension('madmom.ml.hmm', ['madmom/ml/hmm.pyx'], include_dirs=include_dirs),
     Extension(
-        'madmom.ml.nn.layers', ['madmom/ml/nn/layers.py'], include_dirs=include_dirs
+        'madmom.ml.hmm',
+        ['madmom/ml/hmm.pyx'],
+        include_dirs=include_dirs
+    ),
+    Extension(
+        'madmom.ml.nn.layers',
+        ['madmom/ml/nn/layers.py'],
+        include_dirs=include_dirs,
     ),
 ]
 
@@ -44,10 +50,10 @@
 package_data = [
     'models/LICENSE',
     'models/README.rst',
-    'models/beats/201[56]/*',
+    'models/beats/201[569]/*',
     'models/chords/*/*',
     'models/chroma/*/*',
-    'models/downbeats/*/*',
+    'models/downbeats/2016/*',
     'models/key/2018/*',
     'models/notes/*/*',
     'models/onsets/*/*',