anload -- load an annotation item from text description


anload (-f freq) (-t type) (-s|-h|-e|-m|-M|-S) (-H htk.mlf) -T transcription | textfile file


anload constructs a new annotation item in the given file from text descriptions of annotations in a text file. The default format of the text file is one annotation per line. In the normal mode this should have two entries: firstly the time in seconds, secondly the text label (not containing spaces). In the "sample" mode (-s switch), each line should contain three entries: the starting sample number, the final sample number, and the annotation. Ensure that the sampling frequency is correct in the sample mode.


-I Identify the program version.

-f freq Use a sampling interval based on 1/freq seconds for sampling the time of each annotation. Default 10000.

-t type Place a "type" indicator in the created item header, with the argument type.

-s Sample mode. Text lines contain sample numbers or sample ranges, not times. For example: "1000 label" or "1000 2000 label". Ensure sampling frequency is supplied with the -f switch.

-h HTK format load. Equivalent to switches: -s -f 10000000.

-e ESPS compatible label file format load.

-m MBROLA .PHO file compatible label file format load.

-M Millisecond format. Lines contain individual durations in milliseconds (not times) and then label.

-S SAM format label files. Label lines consist of "LBA: <start>, <mid-point>, <end>, <label>" (or with LBB) with extra-long labels continuing with lines labelled "EXT: <label>".

-H htk.mlf Load the annotations from the supplied HTK format master label file. In this case the textfile name refers to the section of the master label file to load. Any scores are stripped during loading.

-T transcription Load the annotation as a single label at time 0 using the supplied argument.


AN Loaded annotation item


1.8 - Mark Huckvale


Except in sample mode, annotation sizes are calculated automatically, the last annotation has length zero.
Fri Jul 09 14:54:30 2004