SpokenLineTracker

Checks a complete spoken attempt against one or more printed lines.

The whole transcript is aligned at once so one misrecognized word does not shift every word that follows it.

quirks, elisions and tokenizer have no default: each comes from the recogniser's or the work's data, and RecognizerQuirks.none is passed by name.

Constructors

constructor(lines: List<String>, quirks: RecognizerQuirks, elisions: Elisions, tokenizer: WordTokenizer)

constructor(line: String, quirks: RecognizerQuirks, elisions: Elisions, tokenizer: WordTokenizer)

data class Progress(val checks: List<WordCheck>)

The result for every expected word in one attempt.

The full forms of the elided spellings the work prints, from the work's data.

Printed words in passage order.

The number of expected words in each printed line.

Model-specific transcription allowances applied while checking.

Splits both the printed lines and every transcript checked against them.

States shown before an attempt: the first word expected and later words ahead.

Pairs the original printed spellings with their check results.

The transcript as its recogniser answers once its hearing table is applied.

Checks a complete recognized transcript against the tracked printed words.

Returns the part of a passage-wide state array belonging to one tracked line.