I need to recognize key words in long speech audio file (wav). There can be different noise, clicks, clangs, etc. This sounds are often recognized as keywords.
How can I set some threshold to distinguish words correctly?
You can not set that. Keyword spotting functionality is not supported by sphinx4.