nryant · nryant · Apr 11, 2020 · Sep 3, 2018 · Sep 3, 2018 · Sep 3, 2018
diff --git a/.github/FUNDING.yml b/.github/FUNDING.yml
@@ -0,0 +1,3 @@
+# These are supported funding model platforms
+
+github: [hbredin]
diff --git a/.github/workflows/doc.yml b/.github/workflows/doc.yml
@@ -0,0 +1,38 @@
+name: Documentation
+on:
+  push:
+    branches:
+      - master
+
+jobs:
+  build-and-deploy:
+    runs-on: ubuntu-latest
+    strategy:
+      max-parallel: 4
+      matrix:
+        python-version: [3.7]
+
+    steps:
+    - uses: actions/checkout@v1
+    - name: Set up Python ${{ matrix.python-version }}
+      uses: actions/setup-python@v1
+      with:
+        python-version: ${{ matrix.python-version }}
+    - name: Install
+      run: |
+        python -m pip install --upgrade pip
+        pip install .
+        pip install -r docs/requirements.txt
+    - name: Build documentation
+      run: |
+        make --directory=docs html
+        touch ./docs/build/html/.nojekyll
+    - name: Deploy
+      env:
+        ACTIONS_DEPLOY_KEY: ${{ secrets.ACTIONS_DEPLOY_KEY }}
+        PUBLISH_BRANCH: gh-pages
+        PUBLISH_DIR: ./docs/build/html
+        SCRIPT_MODE: true
+      run: |
+        wget https://raw.githubusercontent.com/peaceiris/actions-gh-pages/v2/entrypoint.sh
+        bash ./entrypoint.sh
diff --git a/.github/workflows/pypi.yml b/.github/workflows/pypi.yml
@@ -0,0 +1,27 @@
+name: PyPI
+
+on:
+  push:
+    tags:
+    - '*'
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v1
+    - name: Set up Python
+      uses: actions/setup-python@v1
+      with:
+        python-version: '3.x'
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install setuptools wheel twine
+    - name: Build and publish
+      env:
+        TWINE_USERNAME: ${{ secrets.PYPI_USERNAME }}
+        TWINE_PASSWORD: ${{ secrets.PYPI_PASSWORD }}
+      run: |
+        python setup.py sdist bdist_wheel
+        twine upload dist/*
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
@@ -0,0 +1,42 @@
+name: Test
+
+on:
+  pull_request:
+    branches: 
+      - develop
+  push:
+    branches: 
+      - develop
+      - master
+      - release/*
+
+jobs:
+  build:
+
+    runs-on: ubuntu-latest
+    strategy:
+      max-parallel: 4
+      matrix:
+        python-version: [3.6, 3.7]
+
+    steps:
+    - uses: actions/checkout@v1
+    - name: Set up Python ${{ matrix.python-version }}
+      uses: actions/setup-python@v1
+      with:
+        python-version: ${{ matrix.python-version }}
+    - name: Install from source
+      run: |
+        python -m pip install --upgrade pip
+        pip install .
+    - name: Lint with flake8
+      run: |
+        pip install flake8
+        # stop the build if there are Python syntax errors or undefined names
+        flake8 ./pyannote --count --select=E9,F63,F7,F82 --show-source --statistics
+        # exit-zero treats all errors as warnings. The GitHub editor is 127 chars wide
+        flake8 ./pyannote --count --exit-zero --max-complexity=10 --max-line-length=127 --statistics
+    - name: Test with pytest
+      run: |
+        pip install pytest
+        pytest
diff --git a/.travis.yml b/.travis.yml
diff --git a/README.md b/README.md
@@ -1,7 +1,3 @@
-# Announcement
-Open [Phd/postdoc positions](https://mycore.core-cloud.net/public.php?service=files&t=2b5f5a79d24ac81c3b3c371fcd80734b) at [LIMSI](https://www.limsi.fr/en/) combining machine learning, NLP, speech processing, and computer vision. 
-
-
 # pyannote.metrics
 
 > a toolkit for reproducible evaluation, diagnostic, and error analysis of speaker diarization systems

diff --git a/docs/requirements.txt b/docs/requirements.txt
@@ -0,0 +1,3 @@
+Sphinx==2.2.2
+ipython==7.10.1
+sphinx_rtd_theme==0.4.3
diff --git a/docs/source/basics.rst b/docs/source/basics.rst
@@ -38,6 +38,7 @@ As such, they share a common set of methods.
 For instance, once instantiated, they can be called directly to compute the value of the evaluation metric.
 
 .. ipython::
+   :okwarning:
 
    In [10]: from pyannote.metrics.diarization import DiarizationErrorRate
 
@@ -52,6 +53,7 @@ Accumulation & reporting
 The same metric instance can be used to evaluate multiple files.
 
 .. ipython::
+   :okwarning:
 
    In [11]: other_reference = Annotation(uri='file2')
       ....: other_reference[Segment(0, 5)] = 'A'
@@ -116,12 +118,14 @@ For instance, the diarization error rate is the combination of false alarm (non-
 Using ``detailed=True`` will return the value of each component:
 
 .. ipython::
+   :okwarning:
 
    In [13]: metric(reference, hypothesis, detailed=True)
 
 The accumulated value of each component can also be obtained using the overriden :func:`~pyannote.metrics.base.BaseMetric.__getitem__` operator:
 
 .. ipython::
+   :okwarning:
 
    In [13]: metric(other_reference, other_hypothesis)
 

diff --git a/docs/source/changelog.rst b/docs/source/changelog.rst
@@ -2,6 +2,56 @@
 Changelog
 #########
 
+Version 2.3 (2020-02-26)
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+- feat: add DetectionPrecisionRecallFMeasure compound metric (@MarvinLvn)
+- fix: fix corner "in f-measure" case when both precision and recall are zero (@MarvinLvn)
+- fix: fix a typo in documentation (@wq2012)
+
+Version 2.2 (2019-12-13)
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+- feat: add support for evaluation of overlapped speech detection
+- feat: setup continuous integration
+- setup: switch to pyannote.core 3.2
+
+Version 2.1 (2019-06-24)
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+- chore: rewrite mapping and matching routines
+- fix: remove buggy xarray dependency
+- setup: switch to pyannote.core 3.0
+
+Version 2.0.2 (2019-04-15)
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- fix: avoid division by zero
+
+Version 2.0.1 (2019-03-20)
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- BREAKING: drop support for all file formats but RTTM
+- BREAKING: drop Python 2.7 support
+- setup: switch to pyannote.database 2.0
+- setup: switch to pyannote.core 2.1
+
+Version 1.8.1 (2018-11-19)
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- setup: switch to pyannote.core 2.0
+
+Version 1.8 (2018-09-03)
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+- feat: add compound segmentation metric SegmentationPurityCoverageFMeasure (@diego-fustes)
+- fix: fix typo in IdentificationErrorAnalysis (@benjisympa)
+
+Version 1.7.1 (2018-09-03)
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- fix: fix broken images in documentation
+
 Version 1.7 (2018-03-17)
 ~~~~~~~~~~~~~~~~~~~~~~~~
 

diff --git a/docs/source/cli.rst b/docs/source/cli.rst
@@ -13,7 +13,7 @@ Here is an example use of the command line interface that is provided to solve t
 
 .. code-block:: bash
 
-    $ pyannote.metrics.py diarization --subset=development Etape.SpeakerDiarization.TV hypothesis.mdtm
+    $ pyannote.metrics.py diarization --subset=development Etape.SpeakerDiarization.TV hypothesis.rttm
 
     Diarization (collar = 0 ms)               error    purity    coverage     total    correct      %    fa.      %    miss.     %    conf.      %
     --------------------------------------  -------  --------  ----------  --------  ---------  -----  ------  -----  -------  ----  -------  -----
@@ -34,7 +34,7 @@ Tasks
 
 Not only can ``pyannote.metrics.py`` command line tool be used to compute the diarization error rate using NIST implementation, one can also evaluate the typical four sub-modules used in  most speaker diarization systems:
 
-.. image:: images/pipeline.pdf
+.. image:: images/pipeline.png
 
 Practically, the first positional argument (e.g. ``diarization``, above) is a flag indicating which task should be evaluated.
 
@@ -53,8 +53,7 @@ Results are both reported for each file in the selected subset, and aggregated i
 
 As of March 2017, ``pyannote.database`` packages exist for the ETAPE corpus, the REPERE corpus, and the AMI corpus. As more people contribute new ``pyannote.database`` packages, they will be added to the `pyannote` ecosystem.
 
-
 File formats
 ------------
 
-While the MDTM file format is used in this example, several other file formats are available (and can be contributed) thanks to the internal use of the ``pyannote.parser` package.
+Hypothesis files must use the [Rich Transcription Time Marked](https://web.archive.org/web/20170119114252/http://www.itl.nist.gov/iad/mig/tests/rt/2009/docs/rt09-meeting-eval-plan-v2.pdf) (RTTM) format.
diff --git a/docs/source/images/diagnostic.pdf b/docs/source/images/diagnostic.pdf
diff --git a/docs/source/images/diagnostic.png b/docs/source/images/diagnostic.png
diff --git a/docs/source/images/pipeline.pdf b/docs/source/images/pipeline.pdf
diff --git a/docs/source/images/pipeline.png b/docs/source/images/pipeline.png
diff --git a/docs/source/images/segmentation.pdf b/docs/source/images/segmentation.pdf
diff --git a/docs/source/images/segmentation.png b/docs/source/images/segmentation.png
diff --git a/docs/source/reference.rst b/docs/source/reference.rst
@@ -12,7 +12,7 @@ Evaluation metrics
 
 Here is a typical speaker diarization pipeline:
 
-.. image:: images/pipeline.pdf
+.. image:: images/pipeline.png
 
 The first step is usually dedicated to speech activity detection, where the objective is to get rid of all non-speech regions.
 Then, speaker change detection aims at segmenting speech regions into homogeneous segments.
@@ -54,7 +54,7 @@ Segmentation
 
 Change detection modules can be evaluated using two pairs of dual metrics: precision and recall, or purity and coverage.
 
-.. image:: images/segmentation.pdf
+.. image:: images/segmentation.png
 
 Precision and recall are standard metrics based on the number of correctly detected speaker boundaries. Recall is 75% because 3 out of 4 reference boundaries were correctly detected, and precision is 100% because all hypothesized boundaries are correct.
 
@@ -109,7 +109,7 @@ Purity and coverage are two dual evaluation metrics that provide additional insi
    \text{purity} & = & \frac{\displaystyle \sum_{\text{cluster}} \max_{\text{speaker}} |\text{cluster} \cap \text{speaker}|  }{\displaystyle \sum_{\text{cluster}} |\text{cluster}|} \\
   \text{coverage} & = & \frac{\displaystyle \sum_{\text{speaker}} \max_{\text{cluster}} |\text{speaker} \cap \text{cluster}|  }{\displaystyle \sum_{\text{speaker}} |\text{speaker}|} \\
 
-where :math:`|\text{speaker}|` (respectively :math:`|\text{cluster}|` is the speech duration of this particular reference speaker (resp. hypothesized cluster), and  :math:`|\text{speaker} \cap \text{cluster}|` is the duration of their intersection.
+where :math:`|\text{speaker}|` (respectively :math:`|\text{cluster}|`) is the speech duration of this particular reference speaker (resp. hypothesized cluster), and  :math:`|\text{speaker} \cap \text{cluster}|` is the duration of their intersection.
 
 Over-segmented results (e.g. too many speaker clusters) tend to lead to high purity and low coverage, while under-segmented results (e.g. when two speakers are merged into one large cluster) lead to low purity and higher coverage.
 
@@ -119,7 +119,7 @@ Use case
 This figure depicts the evolution of a multi-stage speaker diarization system applied on the ETAPE dataset.
 It is roughly made of four consecutive modules (segmentation, BIC clustering, Viterbi resegmentation, and CLR clustering).
 
-.. image:: images/diagnostic.pdf
+.. image:: images/diagnostic.png
 
 From the upper part of the figure (DER as a function of the module), it is clear that each module improves the output of the previous one.
 

diff --git a/docs/source/tutorial.rst b/docs/source/tutorial.rst
@@ -31,6 +31,7 @@ Several evaluation metrics are available, including the diarization error rate:
 
 
 .. ipython::
+   :okwarning:
 
    In [13]: from pyannote.metrics.diarization import DiarizationErrorRate
 

diff --git a/pyannote/metrics/__init__.py b/pyannote/metrics/__init__.py
@@ -3,7 +3,7 @@
 
 # The MIT License (MIT)
 
-# Copyright (c) 2012-2016 CNRS
+# Copyright (c) 2012-2019 CNRS
 
 # Permission is hereby granted, free of charge, to any person obtaining a copy
 # of this software and associated documentation files (the "Software"), to deal
@@ -26,7 +26,8 @@
 # AUTHORS
 # Hervé BREDIN - http://herve.niderb.fr
 
-from __future__ import unicode_literals
+from multiprocessing import Manager
+from .base import f_measure
 
 from ._version import get_versions
 __version__ = get_versions()['version']
@@ -36,7 +37,7 @@
 # please open a pull request.
 # I had to use this global multiprocessing manager for it to be accessible
 # from anywhere, and make parallel computation of evaluation metric a reality
-from multiprocessing import Manager
 manager_ = Manager()
 
-from .base import f_measure
+
+__all__ = ['f_measure']