Changelog

fix: remove deprecated use of np.NaN

BREAKING: drop support to Python < 3.10

BREAKING: switch to native namespace package

setup: switch to uv

BREAKING: improve diarization purity and coverage to account for overlapping regions

chore: use bool instead of deprecated np.bool

fix: fix corner case for confidence interval

doc: add type hinting (@hadware)

feat: add option to override existing "uri"

feat: add support for missing "uri"

BREAKING: remove (buggy) support for parallel processing

fix: fix documentation deployment

setup: switch to pyannote.database 4.0+

feat: add DetectionCostFunction detection metric (@nryant)

BREAKING: rename pyannote-metrics.py CLI to pyannote-metrics

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)

feat: add support for evaluation of overlapped speech detection

feat: setup continuous integration

setup: switch to pyannote.core 3.2

chore: rewrite mapping and matching routines

fix: remove buggy xarray dependency

setup: switch to pyannote.core 3.0

fix: avoid division by zero

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

setup: switch to pyannote.core 2.0

feat: add compound segmentation metric SegmentationPurityCoverageFMeasure (@diego-fustes)

fix: fix typo in IdentificationErrorAnalysis (@benjisympa)

fix: fix broken images in documentation

feat: add option to filter out target trials in "spotting" mode

chore: default to "parallel=False"

fix: fix Diarization{Purity | Coverage} with empty references

improve: improve support for speaker spotting experiments

chore: (temporarily?) remove parallel processing in pyannote.metrics.py

setup: drop support for Python 2

feat: add fixed vs. variable latency switch for LLSS

fix: add more safety checks to pyannote-metrics.py "spotting" mode

setup: switch to pyannote.core 1.2, pyannote.database 1.1, pyannote.parser 0.7

improve: set latency of missed detections to maximum possible value

improve: improve instructions in pyannote-metrics.py –help

feat: add LowLatencySpeakerSpotting metric

feat: add "spotting" mode to pyannote-metrics.py

setup: switch to pyannote.database 1.0

feat: add "skip_overlap" option to not evaluate overlapping speech regions

improve: bring performance improvement to diarization metrics

fix: fix a bug where collar was applied twice in DiarizationErrorRate

fix: add collar support to purity/coverage/homogeneity/completeness

fix: fix a bug happening in ‘uemify’ when both reference and hypothesis are empty

fix: fix a "division by zero" error in homogeneity/completeness

setup: switch to pyannote.core 1.1 (major performance improvements)

feat: add method DiarizationPurityCoverageFMeasure.compute_metrics to get purity, coverage, and their F-measure (all at once)

feat: add new metric ‘DiarizationPurityCoverageFMeasure’

doc: update installation instructions

setup: switch to pyannote.core 1.0.4

setup: switch to pyannote.core 1.0

feat: add score calibration for binary classification tasks

doc: update citation

doc: update notebook to latest version

doc: add Sphinx documentation

feat: better README and technical report

chore: rename SegmentationError to SegmentationErrorAnalysis

fix: fix DetectionErrorRate support for kwargs

feat: add "parallel" option to not use multiprocessing

feat: add "accuracy" in "detection" report

setup: switch to pyannote.core 0.13

setup: switch to pyannote.parser 0.6.5

feat: add pyannote-metrics.py evaluation script

fix: fix BaseMetric.report() for metric without a ‘total’ component

fix: fix (Greedy)DiarizationErrorRate uem handling

fix: fix (Greedy)DiarizationErrorRate parallel processing

setup: switch to pyannote.core 0.12

setup: update munkres & matplotlib dependencies

feat: support for multiprocessing

feat: add report() method

feat: travis continuous integration (finally!)

improve: speed up detection metrics

feat: add unit tests for detection metrics

fix: fix python 3 support

setup: remove dependency to pyannote.algorithms

setup: switch to pyannote.core 0.11

feat: add pyannote.metrics.binary_classification module

fix: fix (greedy) diarization error rate

feat: add support for ‘collar’ to (greedy) diarization error rate

fix: fix default "xlim" in "plot_distributions"

setup: switch to pyannote.core 0.8 and pyannote.algorithms 0.6.6

feat: add "uem" support to diarization metrics

feat: add plotting functions for binary classification tasks

feat: detection accuracy

refactor: detection metrics

setup: update to pyannote.core 0.7.2

setup: update to pyannote.core 0.6.6

feat: greedy diarization error rate

feat: Python 3 support

feat: unit tests

wip: travis

refactor: diarization metrics

fix: identification error analysis matrix confusion

feat(error): identification regression analysis

feat: new pyannote_eval.py CLI

feat(error): segmentation error analysis

feat(detection): add precision and recall

fix(identification): fix precision and recall

feat(segmentation): add precision and recall

feat(identification): add support for NIST collar

feat(error): add module for detailed error analysis

first public version