#include <srch.h>
◆ adapt_am
◆ ascale
Same as senscale but it records the senscale for the whole sentence. The default size is 3000 frames.
◆ ascale_sz
◆ ascr
Pointer to Senone and composite senone scores for one frame
◆ beam
Pointer to Structure that wraps up parameters related to beam pruning
◆ cache_win
The windows lengths of the cache for approximate search
◆ cache_win_strt
int32 srch_s::cache_win_strt |
The start index of the window near the end of a block
◆ dag
Word graph representation of completed utterance.
◆ exit_id
Lattice ID for exit word (-1 if utterance is not complete)
◆ fastgmm
Pointer to Structure that wraps up parameters for fast GMM computation
◆ funcs
Methods specific to a particular search mode.
◆ grh
Pointer to search specific structures
◆ hmm_dump_ef
int32 srch_s::hmm_dump_ef |
End frame for HMMs to be dumped for debugging
◆ hmm_dump_sf
int32 srch_s::hmm_dump_sf |
Start frame for HMMs to be dumped for debugging
◆ hmmdumpfp
Copy of File handle for dumping hmms for debugging
◆ kbc
Pointer to AM adaptation structure Pointer to the kbcore
◆ matchfp
Copy of File handle for the match file
◆ matchsegfp
Copy of File handle for the match segmentation file
◆ num_frm
Number of frames processed
◆ num_segs
In one search, (i.e. between one srch_utt_begin and one srch_utt_end) The number of segments the search has decoded. That also means the number of times srch_utt_decode_blk is called.
◆ op_mode
◆ pl
Pointer to Structure that wraps up parameter for phoneme look-ahead
◆ segsz
Size of segments for each call
◆ segsz_sz
◆ senscale
TEMPORARY VARIABLE: Senone scale
◆ stat
Pointer to the statistics structure
◆ uttfile
Reference to Utterance filename (or NULL)
◆ uttid
The documentation for this struct was generated from the following file:
- /builddir/build/BUILD/sphinx3-0.8/include/srch.h