This seems to be a shortcoming of the MIST design, I talk about it some more in this thread: viewtopic.php?f=115&t=32998&p=339804#p339804
It's a shame, because with just a few extra I/O pins from the FPGA it could be an amazing little sequencer box.
One possibility I considered is to use STE DMA sound and a fancy audio-to-MIDI converter unit. Whether this will work depends on so many factors. It might also be possible to design some kind of device that uses spare pixels on the VGA port to output MIDI data.