I think your driver transistors are way too wimpy. Take a look at the gain curve versus current for the BC337: http://www.ortodoxism.ro/datasheets/philips/BC337_3.pdf
For a 1 amp peak load with something like 10 ma from the opamp, you need a gain of 100, at least. The gain of the BC337 starts to poop out past 200 mA. You can never try to operate transistors at the current given in the max. spec. Always look at the gain curve to see what they are really capable of. You also probably have a power problem with such small devices. At the 1 amp peaks, you have about 5 volts across the transistor, which means at that moment they are producing 5 watts of heat. TO92 packs won't last long under those conditions (it is rated for well under a watt), even if you find one that will carry the current. You could parallel 4 or 5 transistors to get them operating in their high gain region, but you would probably need a small emitter resistor for each to balance them.
There are some TO-92's that will deliver the gain and current, if you can keep them cool, like the Zetex ZTX869, minimum gain of 300 at 1 amp and lots of current capability above that.
But for 30 Hz, where the opamp has lots of time to get through the crossover voltage (getting one half of the follower turned off and the other half turned on), you might consider a couple of 3 (or more) amp darlington transistors
say, TIP120 and TIP125, with their gain curves that peak just above 1 amp:
, with high frequency feedback (small capacitor) around the opamp for stability, but DC feedback after the output of the darlingtons (for low frequency accuracy). Something in a TO-220 package would stand the heat batter than TO-92's.