The sound file could be done by inserting 300ms of silence at the start of the sound clip (in a sound editor like Audacity) and then playing it before the main stimulus but with a Time value of 0ms so that it (i.e., the 300ms of silence) starts in synch with the main stim.

As far as I know you can not send a TTL out in the same way. Once you start recording the RT you can't start another event like that. The only way I could see doing that is to do the sound as described above and send a TTL signal along with it at onset (before the main stim) and give it a 300ms duration. So you wouldn't get the signal at 300ms, but you would get the END of the signal at 300ms. Would that help as a marker?