I don't understand this statement by Brian Armstrong.
Surely if the hashrate drops off, the difficulty just drops to compensate and we continue to generate blocks at the same intervals on average. Is he suggesting that the hashing level can somehow have an impact on the emission rate ?

Difficulty takes approximately 2 weeks to adjust. Currently, miners are pushing out blocks approximately every 10 minutes. If after the halving day 50% of the hashrate drops of the network, only 50% will be trying to solve blocks with the same difficulty. Therefore, it will take miners twice as long to find a block (making blocktimes essentially 20 minutes for a while, which also results in a lower emission). The difficulty adjustment is meant so the blocktime averages to 10 minute on the long term. In case the hashrate shoots up and difficulty hasn't adjusted you will see faster blocks (I think we had average 8 minute blocktimes for a week or so).
P.S. The difficulty adjusts after N number of blocks, thus in above scenario it could take 4 weeks to adjust. There is also a doom scenario where the difficulty will never adjust because to many miners drop off the network.