The required timeslices are quite a bit smaller than 1 second, so one can try to see if the different tasks can complete their work within 1 second.
Voice runs every 5 ms. 1/0.005 = 200 times/second.
Video runs at 25 frames/second. = 25 times/second
Voice needs 1 ms each time it's run = 200 ms/second.
Video needs 20 ms each time it's run = 25*0.020 = 500 ms
2 voice tasks + 1 video task = 200ms*2 + 500ms = 900ms
How one wants an RTOS to schedule such tasks depends among other things on how much jitter you can live with for the different tasks . e.g. the 2 voice tasks could have equal priority, but higher than than the video task - allowing the voice tasks to run whenever they need in a fifo order. (meaning one voice task might need to wait at most 1ms to be scheduled), and the video task gets the remaining CPU time