If you think the measurement distance was accurate then removing the same amount of fixed time from both sets of measurements should give you what you need. If this is what you did already you should be good. The horn data will be the most sensitive to removing the fixed delay and if it isn't right it will show in the phase response.