enough to cover odontocete clicks. The number of samples saved per snippet is determined by the
sum of the pre- and post-trigger times. With the maximum settings, 1728 samples will be saved per
snippet.
4.1.4
Output File formats
The ST click detector produces two additional output files per recording with suffixes 'bcl' and 'dwv'.
These are in addition to the .wav and .xml files generated normally. The Pamguard interface
automatically interprets these files when displaying detection and continuous recording results. For
users interested in accessing the data using other interfaces, the details of the file formats are given
below:
•
xml
This is the metadata file produced by SoundTrap to document each recording. The
ST detector adds metadata about the detector and the user settings to this file.
•
wav
This is a Microsoft WAV format file containing the decimated continuous sound
recording produced in parallel with the detector. The sampling rate is selected by the user
but would typically be 48kHz or 96kHz to maximize recording time. By default, sound data is
compressed in the SoundTrap using the lossless X3 compressor but this is transparent to the
user: the .wav file contains the reconstituted uncompressed data.
•
dwv
This is a Microsoft WAV format file containing the wideband sound snippets for each
detection. The sampling rate is 576 kHz and the snippets are arranged contiguously with a
constant number of samples per snippet. The size of each snippet is set the by the pre- and
post-trigger user parameters. The snippet size in samples will be close to
round((pre+post)*576e3) but the precise sample count can be found in the .xml file (see the
Matlab tools for an example of how to extract this information automatically). By default,
snippet sound data is compressed using the lossless X3 compressor but this is transparent to
the user: the .dwv file contains the reconstituted uncompressed data. Up to 100000 snippets
will be saved per recording (i.e., per output file). If more detections are made over this
interval, they will be reported in the BCL file (see below) but no snippet will be saved. This
puts an upper limit on the memory usage rate making it easier to calculate for how long the
SoundTrap will be able to record.
•
bcl
This is a comma-separated-variable (CSV) text file containing information about
detections. Each line has 7 columns with the following format:
Seconds (in Unix seconds), Microseconds, Report type, Info1,Info2,Info3,Info4,
