RESPITE: The CASA Toolkit Page: Documentation: Block Library Index:DeltaBounds |

DeltaBounds

The ** DeltaBounds** block can be used for calculating missing data `bounds constraints' for delta features. As input it takes a set of static features, a missing data mask for the static features and lower and upper bounds for the static features. As output it generates delta features, a missing data mask for the deltas features and sensible lower and upper bounds for the delta features.

The static features are converted to delta features using the same procedure as employed by the ** Delta** block.

The static missing data mask is converted into a missing data mask for delta features using the same procedure as the ** DeltaMask** block.

The delta bounds are computed by considering the largest and small delta values that can be achieved given the bounds on the missing static features. So to calculate the upper delta bound, the left hand side of the delta regression window (see ** Delta** block) takes the smallest allowed values of the missing static features, and the right hand side takes the largest allowed values. For the lower delta bound the left side takes large values and the right side takes small values.

Note, the static mask and bounds inputs are optional. If the static bounds are not supplied, the delta features and mask will still be computed, but the delta bounds will not be computed and no data will arrive at the delta bounds outputs. Likewise, if the static mask is not supplied, then neither the delta mask nor the delta bounds can be computed and only the delta feature output will be active.

Inputs |
Meaning | Sample | 1-D frame | 2-D frame |
---|---|---|---|---|

in1 |
static features | No | Yes | No |

(in2) |
mask frame | No | Yes | No |

(in3) |
lower bounds | No | Yes | No |

(in4) |
upper bounds | No | Yes | No |

Outputs |
Meaning |
---|---|

out1 |
delta features |

out2 |
mask frame for delta features |

out1 |
lower bounds for delta features |

out1 |
upper bounds for delta features |

Parameters |
Type | Default | Meaning |
---|---|---|---|

DELTA_WIN |
Integer | 2 | Half size of the regression window |

Documentation for CTKv1.1.4 - Last modified: Wed Jun 27 17:11:06 BST 2001