|
KeenASR Framework v2.2 (c662ac9)
Keen Research
|
Result of aligning a recognized token sequence against a reference. More...
#include <KIOSTextAligner.h>
Properties | |
| NSArray< KIOSAlignmentItem * > * | trace |
| Full alignment trace, left-to-right over the input sequences. | |
| NSInteger | matches |
| Number of matched token pairs. | |
| NSInteger | substitutions |
| Number of substitutions in the alignment. | |
| NSInteger | insertions |
| Number of insertions (extra recognized tokens). | |
| NSInteger | deletions |
| Number of deletions (skipped reference tokens). | |
| NSInteger | refLength |
| Length of the post-normalization reference token vector. | |
| NSInteger | recLength |
| Length of the post-normalization recognized token vector. | |
| float | wordErrorRate |
| Word error rate: (substitutions + insertions + deletions) / refLength. | |
| NSArray< NSNumber * > * | matchedRefIndices |
| Sorted array of reference indices that matched their corresponding recognized token (NSNumber-wrapped NSInteger). | |
| NSArray< NSNumber * > * | skippedRefIndices |
| Sorted array of reference indices that did not match but lie at or below furthestMatchedIndex (NSNumber-wrapped NSInteger). | |
| NSInteger | furthestMatchedIndex |
| Highest matched reference index, or NSNotFound if no reference token matched. | |
| NSArray< NSNumber * > * | repetitionRefIndices |
| Reference indices flagged as repetitions in the recognized text (NSNumber-wrapped NSInteger). | |
Result of aligning a recognized token sequence against a reference.
Exposes both an edit-script trace (suitable for word-error-rate reporting) and derived oral-reading views (matched / skipped / furthest-reached).
Indices in all *RefIndices accessors and furthestMatchedIndex refer to positions in the normalized reference token vector, accessible via KIOSTextAligner.referenceTokens.
|
readnonatomicassign |
Sorted array of reference indices that matched their corresponding recognized token (NSNumber-wrapped NSInteger).
Substitutions and deletions are not counted as matched on the reference side — only exact-word matches.
|
readnonatomicassign |
Reference indices flagged as repetitions in the recognized text (NSNumber-wrapped NSInteger).
Populated only when KIOSAlignmentConfig.detectRepetitions was YES on the call that produced this result. Empty otherwise.
|
readnonatomicassign |
Sorted array of reference indices that did not match but lie at or below furthestMatchedIndex (NSNumber-wrapped NSInteger).
Reference tokens past the furthest-matched index are treated as "not yet reached" rather than skipped, and are excluded here. Useful for read-along UIs that highlight skipped words.
|
readnonatomicassign |
Word error rate: (substitutions + insertions + deletions) / refLength.
Returns 0.0 when refLength is 0.