I use PocketSphinx_continues.exe in grammar mode, with a small grammar (12 sentences, some phrases, some isolated words). Frequently this happens: I say one of the words and repeat saying it many times. It recognize it several times, and then not recognize it several times (it replaces it with another word in the grammaer, which actually sounds very different), and then it recognize it again. I pronounce the word exectly the same all the time.
The same happens when I say a specific sentence several times - it doesn't recognize one of the words (mostly the first word). It replaces it with another sentence in the grammar, which is identical except this word.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I use PocketSphinx_continues.exe in grammar mode, with a small grammar (12 sentences, some phrases, some isolated words). Frequently this happens: I say one of the words and repeat saying it many times. It recognize it several times, and then not recognize it several times (it replaces it with another word in the grammaer, which actually sounds very different), and then it recognize it again. I pronounce the word exectly the same all the time.
The same happens when I say a specific sentence several times - it doesn't recognize one of the words (mostly the first word). It replaces it with another sentence in the grammar, which is identical except this word.
You need to us a means to reproduce your problem. You need to provide the audio file, command line you are using, models and logs.