it also cannot detect if someone is speaking background and mistake as input.... anyone can solve this?