Are you sending audio over this leased line? The speech recognition engines VoiceAttack utilizes do not require an internet connection to function, if that's what you're referring to. Recognition is performed on the machine VoiceAttack is installed on.
Background noise in general can disrupt recognition, if your microphone picks it up. The speech recognition engines used are intended for a quiet environment.
You could try using a third-party application to filter out certain background noises, however filtering in itself can also negatively affect speech recognition accuracy, if (part of) the data the engine relies on is also filtered out.
If you are using the SAPI speech recognition engine (if you didn't manually install a speech recognition engine, that would be what you're using), you'd want to make sure to perform the speech recognition training at least three times (instructions for starting the training can be found
in this topic).
There is another engine available, which does not need to be/cannot be trained, and has been reported by some users to work better for them in their specific circumstances:
Speech Platform 11Note that this engine is specifically designed for predefined commands (as you would normally create using VoiceAttack), and as such does not have dictation capabilities.