sample.beatsdrop.decision module

Decision models for discriminating beating trajectories

class sample.beatsdrop.decision.AlmostNotABeatDecisor(th: float = 1, intermediate: Optional[OptionalStorage] = None, **kwargs)

Bases: BeatDecisor

Decide if a trajectory is a beat or not based on the fact that the two amplitudes are not almost zero and the two frequencies are not almost the same.

Parameters:

  • th (float) – Inexact equality threshold, as a multiplier of floating point epsilon.

  • intermediate (OptionalStorage) – Optionally-activatable storage

decide_beat(i: int, t: ndarray, track: dict, beatsdrop: BeatRegression, params: Tuple[float, float, float]) → bool

Decision function based on normalized power square difference

Parameters:

  • i (int) – Track index

  • t (array) – Time axis

  • track (dict) – Track

  • beatsdrop (BeatRegression) – Beat regression model

  • params (triplet of floats) – Frequency, decay, and amplitude

Returns:

True iff the correlation is not significant or is below the threshold

Return type:

bool

class sample.beatsdrop.decision.BeatDecisor(intermediate: Optional[OptionalStorage] = None, **kwargs)

Bases: BaseEstimator

Model responsible for deciding wether the trajectory is a beat or not

decide_beat(i: int, t: ndarray, track: dict, beatsdrop: BeatRegression, params: Tuple[float, float, float]) → bool

Decision function. Should be overwritten by child classes. This method always decides for the single-partial parameters (False) and never for the beats (True)

Parameters:

  • i (int) – Track index

  • t (array) – Time axis

  • track (dict) – Track

  • beatsdrop (BeatRegression) – Beat regression model

  • params (triplet of floats) – Frequency, decay, and amplitude

Returns:

Decision

Return type:

bool

property intermediate

Optionally-activatable storage

output_beat_params(beat_params: Tuple[float, float, float, float, float, float, float, float]) → Sequence[Tuple[float, float, float, float]]

Format beat parameters as output for the sample.sample.SAMPLE._fit_track() method and phases

Parameters:

beat_params (octuplet of floats) – a0, a1, f0, f1, d0, d1, p0, p1

Returns:

Parameters for sample.sample.SAMPLE._fit_track() and phase

output_params(params: Tuple[float, float, float]) → Sequence[Tuple[float, float, float, float]]

Format single-partial parameters as output for the sample.sample.SAMPLE._fit_track() method and phases

Parameters:

params (triplet of floats) – Frequency, decay, and amplitude

Returns:

Parameters for sample.sample.SAMPLE._fit_track() and phase

track_params(i: int, t: ndarray, track: dict, beatsdrop: BeatRegression, params: Tuple[float, float, float], fit: bool = False) → Sequence[Tuple[float, float, float, float]]

Returns linear or beat parameters depending on the decision

Parameters:

  • i (int) – Track index

  • t (array) – Time axis

  • track (dict) – Track

  • beatsdrop (BeatRegression) – Beat regression model

  • params (triplet of floats) – Frequency, decay, and amplitude

  • fit (bool) – If True, then fit the beatsdrop model

Returns:

Parameters for the sample.sample.SAMPLE._fit_track() method