Label statements define different label types for each level in the hierarchy. A level in the hierarchy represents a kind of segment or event which can be marked in the speech signal. Each of these segments or events might have a number of labels: for example, a word level segment may have a Word label for a content/function label and a Text label for the text of the word.
By default, every level in the hierarchy has one label type defined with the same name as the level. If more than one label type is required for a level, they can be defined in a label statement; the syntax of a label statement is:
where levelname is the name of the level for which a label is being defined and labelname is the name of the label associated with this level.label levelname labelname ?number?
The optional number is for compatibility with the older MU+ system and should be the same as the number associated with the label in the labed_template file. This number is only used when converting old .exlab files into Emu format. It is not needed once conversion has been done or in new databases.
For a given utterance, a set of labels associated with a level can be derived in a number of ways, either by reading an external file which contains labels and times (see section Labfile Definitions), automatically using a script or by hand using the Emu labeller.
Example:
label Word Accent label Word Text label Syllable Pitch_Accent |