tempo_eval.Metric

class tempo_eval.Metric(name, formatted_name=None, description=None, eval_function=None, extract_function=None, suitability_function=<function Metric.<lambda>>, significant_difference_function=None, best_value=1.0, signed=False, unit='%')[source]

Metric.

Structured collection of logic and injected functions for different metrics.

Example

>>> from tempo_eval import OE1, read_reference_annotations, read_estimate_annotations
>>> gt_ref = read_reference_annotations('giantsteps_tempo', validate=False)
>>> gt_est = read_estimate_annotations('giantsteps_tempo', validate=False)
# evaluate estimates using the reference values and Metric OE1:
>>> res = OE1.eval_annotations(gt_ref['tempo'], gt_est['tempo'])
# show result of ref '1.0' and est 'davies2009/mirex_qm_tempotracker'
# for file '3453642.LOFI.jams':
>>> res['1.0']['davies2009/mirex_qm_tempotracker']['3453642.LOFI.jams']
[-0.02271693594286862]

__init__(name, formatted_name=None, description=None, eval_function=None, extract_function=None, suitability_function=<function Metric.<lambda>>, significant_difference_function=None, best_value=1.0, signed=False, unit='%')[source]

Create a metric using the given functions.

Parameters

name (str) – name
description (Optional[str]) – HTML-formatted, high level description
formatted_name (Optional[str]) – HTML-formatted name
eval_function (Optional[Callable[[Union[Any, Tuple[float, float, float], float], Union[Any, Tuple[float, float, float], float]], Any]]) – function that compares two tempi, may also accept a tolerance parameter, e.g. equal1.
extract_function (Optional[Callable[[Annotation], Union[Any, Tuple[float, float, float], float]]]) – function to extract tempo values from annotation, e.g. extract_tempo.
significant_difference_function (Optional[Callable[[Dict[str, List[Any]], Dict[str, List[Any]], str, str], List[float]]]) – function to determine significant differences. E.g. mcnemar.
suitability_function (Callable[[Union[Any, Tuple[float, float, float], float]], bool]) – function to determine whether a tempo value is suitable for this metric, e.g. is_single_bpm.
best_value (float) – best value (to show, if no values are available)
unit (Optional[str]) – unit, e.g., '%'
signed (bool) – is the metric signed (e.g. percentage error) or absolute (e.g. Accuracy1)?

Methods

`__init__`(name[, formatted_name, ...])	Create a metric using the given functions.
`are_tempi_suitable`(tempi)	Is at least one tempo meant for this metric?
`averages`(eval_results[, item_id_filter, ...])	Calculate means and standard deviations for the given evaluation results.
`eval_annotations`(reference_annotation_set, ...)	Evaluates annotations.
`eval_tempi`(reference_tempi, estimated_tempi)	Evaluates tempi for all provided tempi for each track.
`extract_tempi`(annotation_set)	Extracts tempi from the given annotation set.
`is_tempo_suitable`(tempo)	Indicates whether a given tempo is suitable for this metric.