This amplifier must crawl before it can walk. YourYour idle current is not well defined ,thermal thermal runaway is possible .The The bottom npn darlingtonNPN Darlington is not being driven in any way so it is trying to be a constant source .This This means that the max efficiency if all the parts were ideal would be 25% instead of the normal 50% max.your Your 680k seems really high so such a high prospective gain means little or no feedback .If If you used say a 68k then there would of course be much less gain but lower distortion.