iv_scoring

Copyright 2014-2019 Anthony Larcher and Sylvain Meignier

iv_scoring provides methods to compare i-vectors

iv_scoring.PLDA_scoring(enroll, test, ndx, mu, F, G, Sigma, test_uncertainty=None, Vtrans=None, p_known=0.0, scaling_factor=1.0, full_model=False)[source]

Compute the PLDA scores between two sets of vectors. The list of trials to perform is given in an Ndx object. PLDA matrices have to be pre-computed. i-vectors are supposed to be whitened before.

Implements the approach described in [Lee13] including scoring for partially open-set identification

Parameters

  • enroll – a StatServer in which stat1 are i-vectors

  • test – a StatServer in which stat1 are i-vectors

  • ndx – an Ndx object defining the list of trials to perform

  • mu – the mean vector of the PLDA gaussian

  • F – the between-class co-variance matrix of the PLDA

  • G – the within-class co-variance matrix of the PLDA

  • Sigma – the residual covariance matrix

  • p_known – probability of having a known speaker for open-set identification case (=1 for the verification task and =0 for the closed-set case)

  • scaling_factor – scaling factor to be multiplied by the sufficient statistics

  • full_model – boolean, set to True when using a complete PLDA model (including within class covariance matrix)

Returns

a score object

iv_scoring.PLDA_scoring_uncertainty(enroll, test, ndx, mu, F, Sigma, p_known=0.0, scaling_factor=1.0, test_uncertainty=None, Vtrans=None, check_missing=True)[source]
Parameters

  • enroll –

  • test –

  • ndx –

  • mu –

  • F –

  • Sigma –

  • p_known –

  • scaling_factor –

  • test_uncertainty –

  • Vtrans –

  • check_missing –

Returns

iv_scoring.cosine_scoring(enroll, test, ndx, wccn=None, check_missing=True)[source]

Compute the cosine similarities between to sets of vectors. The list of trials to perform is given in an Ndx object.

Parameters

  • enroll – a StatServer in which stat1 are i-vectors

  • test – a StatServer in which stat1 are i-vectors

  • ndx – an Ndx object defining the list of trials to perform

  • wccn – numpy.ndarray, if provided, the i-vectors are normalized by using a Within Class Covariance Matrix

  • check_missing – boolean, if True, check that all models and segments exist

Returns

a score object

iv_scoring.fast_PLDA_scoring(enroll, test, ndx, mu, F, Sigma, test_uncertainty=None, Vtrans=None, p_known=0.0, scaling_factor=1.0, check_missing=True)[source]

Compute the PLDA scores between to sets of vectors. The list of trials to perform is given in an Ndx object. PLDA matrices have to be pre-computed. i-vectors are supposed to be whitened before.

Parameters

  • enroll – a StatServer in which stat1 are i-vectors

  • test – a StatServer in which stat1 are i-vectors

  • ndx – an Ndx object defining the list of trials to perform

  • mu – the mean vector of the PLDA gaussian

  • F – the between-class co-variance matrix of the PLDA

  • Sigma – the residual covariance matrix

  • p_known – probability of having a known speaker for open-set identification case (=1 for the verification task and =0 for the closed-set case)

  • check_missing – boolean, if True, check that all models and segments exist

Returns

a score object

iv_scoring.full_PLDA_scoring(enroll, test, ndx, mu, F, G, Sigma, p_known=0.0, scaling_factor=1.0, check_missing=True)[source]

Compute PLDA scoring

Parameters

  • enroll – a StatServer in which stat1 are i-vectors

  • test – a StatServer in which stat1 are i-vectors

  • ndx – an Ndx object defining the list of trials to perform

  • mu – the mean vector of the PLDA gaussian

  • F – the between-class co-variance matrix of the PLDA

  • G – the within-class co-variance matrix of the PLDA

  • Sigma – the residual covariance matrix

  • p_known – probability of having a known speaker for open-set identification case (=1 for the verification task and =0 for the closed-set case)

  • check_missing – boolean, default is True, set to False not to check missing models

iv_scoring.mahalanobis_scoring(enroll, test, ndx, m, check_missing=True)[source]

Compute the mahalanobis distance between to sets of vectors. The list of trials to perform is given in an Ndx object.

Parameters

  • enroll – a StatServer in which stat1 are i-vectors

  • test – a StatServer in which stat1 are i-vectors

  • ndx – an Ndx object defining the list of trials to perform

  • m – mahalanobis matrix as a ndarray

  • check_missing – boolean, default is True, set to False not to check missing models

Returns

a score object

iv_scoring.two_covariance_scoring(enroll, test, ndx, W, B, check_missing=True)[source]

Compute the 2-covariance scores between to sets of vectors. The list of trials to perform is given in an Ndx object. Within and between class co-variance matrices have to be pre-computed.

Parameters

  • enroll – a StatServer in which stat1 are i-vectors

  • test – a StatServer in which stat1 are i-vectors

  • ndx – an Ndx object defining the list of trials to perform

  • W – the within-class co-variance matrix to consider

  • B – the between-class co-variance matrix to consider

  • check_missing – boolean, default is True, set to False not to check missing models

Returns

a score object