Transcripts which have been evaluated against an ASR result are aligned, token by token. For a given scoring run in tblLVST_ScoringProtocol, this table tracks the aligned tokens. Note that the transcript and SRE are tokens defined by the grammar, not necessarily single words. For example: TRANSCRIPT: my dog buck ASR: dog buick The grammar may have 'my dog' as one token, and 'buck' or 'buick' as a single token. One possible alignment would be: TRANSCRIPT ASR my dog dog buck buick In the case of semantic or grammatical scoring, the transcript token will be either 'YES' or 'NO', and the ASR output token will always be 'YES'; comparing 'YES' to 'YES' or 'NO' to 'YES' indicates whether the semantic or grammar comparisons were successful.
| Column Name | Data Type | Data Length | Default Value | See Also |
|
TranscriptsAlignedID |
INTEGER |
N/A |
N/A |
N/A |
|
ScoringProtocolID |
INTEGER |
N/A |
N/A |
N/A |
|
Sequence |
INTEGER |
N/A |
0 |
N/A |
|
Confidence |
FLOAT |
N/A |
0 |
N/A |
|
stTranscript |
INTEGER |
N/A |
N/A |
|
|
stSREOutput |
INTEGER |
N/A |
N/A |
|
|
cInVocabulary |
INTEGER |
N/A |
-1 |
N/A |
|
cScoreType |
INTEGER |
N/A |
|
N/A |
TranscriptsAlignedID
| Foreign key | Reference table | Reference key |
|
ScoringProtocolID |
tblLVST_ScoringProtocol |
ScoringProtocolID |
| Foreign key | Reference table | Reference key |
|
stTranscript |
tblLVST_StringTable |
StringID |
| Foreign key | Reference table | Reference key |
|
stSREOutput |
tblLVST_StringTable |
StringID |
| Foreign key | Reference table | Reference key |
|
cInVocabulary |
tblLVST_TranscriptsAligned_cInVocabulary |
ConstantValue |
| Foreign key | Reference table | Reference key |
|
cScoreType |
tblLVST_TranscriptsAligned_cScoreType |
ConstantValue |
TranscriptsAlignedID
Description:Autoincrementing primary key.
Constraint Name:N/A
Constraint Description:N/A
ScoringProtocolID
Description:The scoring protocol used to derive this alignment. Foreign key to tblLVST_ScoringProtocol.
Constraint Name:N/A
Constraint Description:N/A
Sequence
Description:The linear order of this alignment, starting at 0 and increasing. For example: 'my dog buck' TOKEN SEQUENCE my 0 dog 1 buck 2
Constraint Name:N/A
Constraint Description:N/A
Confidence
Description:The confidence for the SREOutput. Note that if the SREOutput token is empty (indicating a DELETION), then there won't be a confidence score. Remember that when working with the confidence scores.
Constraint Name:N/A
Constraint Description:N/A
stTranscript
Description:Foreign key to tblLVST_StringTable. The transcript token used in this alignment.
Constraint Name:N/A
Constraint Description:
stSREOutput
Description:Foreign key to tblLVST_StringTable. The output token from the ASR engine.
Constraint Name:N/A
Constraint Description:N/A
cInVocabulary
Description:Foreign key to tblLVST_TranscriptsAligned_cInVocabalary. Indicates whether the token is in-vocabulary or not. In-vocabulary measurements indicate whether the token is somewhere in the grammar, regardless of whether the transcript actually parses in the grammar or not. In other words, treat all the terminal tokens in a grammar as a bag, and see if this token is in that bag. This allows you to make decisions which indicate whether at least callers are using the right words, if not the correct syntax, nearly always of secondary importance to whether the transcript is In or Out of Grammar. Maps to the ConstantValue indicated by by 'IN_VOCABULARY', 'OUT_OF_VOCABULARY', 'UNKNOWN'.
Constraint Name:'IN_VOCABULARY', 'OUT_OF_VOCABULARY', 'UNKNOWN
Constraint Description:Foreign key to tblLVST_TranscriptsAligned_cInVocabalary.
cScoreType
Description:Foreign key to tblLVST_TranscriptsAligned_cScoreType. Marks this alignment as being one of the 8 score types in tblLVST_TranscriptsAligned_cScoreType.
Constraint Name:ScoreTypeMustExist
Constraint Description:Foreign key to tblLVST_TranscriptsAligned_cInVocabalary.
Constraint Name:N/A
Constraint Description:N/A
Columns related:N/A