|
libak
0.4.0
|
Word graph edge. More...
Data Fields | |
| unsigned long | w |
| Encoded Word of the transition. | |
| akProb | a |
| Log-probability related to the acoustic (and lexicon) models. | |
| akProb | l |
| Log-probability related to the language model. | |
| akWGState * | from |
| Source state. | |
| akWGState * | to |
| Destination state. | |
| akWGSegment * | segs |
| Word segmentation, can be NULL. | |
| size_t | nsegs |
| Number of segments. | |
Word graph edge.
This is the structure used to represent the edges in a word graph.
Log-probability related to the acoustic (and lexicon) models.
Source state.
Log-probability related to the language model.
| size_t akWGEdge::nsegs |
Number of segments.
Word segmentation, can be NULL.
Destination state.
| unsigned long akWGEdge::w |
Encoded Word of the transition.
A value equal to the size of the dictionary of words used to generate the word graph is interpreted as the especial final word. ULONG_MAX denotes the especial lambda (empty) word.
1.7.6.1