Can you clarify this issue for me a bit? I would like to help more, but it is dependent on it being very clear about what you are experiencing. Which of the following describes the issue:
• There is silence and recognition is frequently occurring.
• There is noise in the environment and when the app hears the noise, it is giving a hypothesis which is a word in the vocabulary.
• There is human speech in the environment and when the app hears speech it is giving a hypothesis which is a word in the vocabulary, but the speech was not any of the words in the vocabulary.