構造体 WORD_INFO
Word dictionary structure to hold vocabulary. [詳細]
#include <vocabulary.h>
WORD_INFOのコラボレーション図
変数 | |
| WORD_ID | maxnum |
| Allocated number of word space. | |
| WORD_ID | num |
| Number of words. | |
| WORD_ID | errnum |
| Number of error words that were skipped when reading dictionary. | |
| unsigned char * | wlen |
| Number of phonemes for each word [wid]. | |
| char ** | wname |
| Word name string for each word [wid]. With DFA, it's category ID. With N-gram, it's N-gram entry name. | |
| char ** | woutput |
| Word output string that will be output as recognition result for each word [wid]. | |
| HMM_Logical *** | wseq |
| Phone sequence of each word [wid][0..wlen[wid]-1]. | |
| WORD_ID * | wton |
| Reference to N-gram/category ID of each word ID [wid]. | |
| LOGPROB * | cprob |
| Class probability of each word [wid]. | |
| WORD_ID | cwnum |
| Number of words whose class prob is specified (just for a statistic). | |
| WORD_ID | head_silwid |
| Word ID of beginning-of-sentence silence. | |
| WORD_ID | tail_silwid |
| Word ID of end-of-sentence silence. | |
| short | maxwn |
| Maximum number of HMM states per word (statistic). | |
| short | maxwlen |
| Maximum number of phones in a word (statistic). | |
| boolean * | is_transparent |
| TRUE if the word can be treated as transparent [wid]. | |
| APATNODE * | errph_root |
| Root node of index tree for gathering error HMM name appeared when reading the dictionary. | |
| BMALLOC_BASE * | mroot |
| Pointer for block memory allocation. | |
説明
Word dictionary structure to hold vocabulary.
vocabulary.h の 57 行で定義されています。
この構造体の説明は次のファイルから生成されました:
- libsent/include/sent/vocabulary.h
