Next: Check List Post-processing
Up: Post-processing
Previous: Special Conversion for Annotation
Contents
Automatic Error Detection
Automatic error detection denotes all check procedures that may be carried out
automatically. There is no certain place in the chain of the post-processing
where automatic error checks are most recommended. Whenever possible include
automatic checks after each processing step, especially after steps that
require manual work.
Here are some possible checks that can be carried out automatically:
- Check for empty files
- Check for correct length
- Check for correct terminology
- Check for speech contained in the file or only silence
- Check for required number of files per recording session
- Check for total disk space of a recording session
- Check for S/N ratio
- Check for consistent header content
An easy way to implement such automatic checks is the usage of a UNIX shell.
Most checks are simple UNIX commands or can be easily implemented in a script
language like AWK, perl or BASH. If you are using MS platforms7.9, you can use the public domain
Cygwin applications that allow you to
run a UNIX-like environment on your PC7.10.
You may check [11], chapter 5 for a collection of small examples scripts.
In a large data pipeline it may happen that checks are omitted and nobody
notices until later. A good way to avoid this is a central log file or database
where all checker programs may enter their results together with date and the
checked data.
Next: Check List Post-processing
Up: Post-processing
Previous: Special Conversion for Annotation
Contents
BITS Projekt-Account
2004-06-01