What do you expect? There is no magic flag to increase the speed and no better function, which is not mentioned in the documentation.
Is your function to be integrated vectorized already? Did you reduce the number of expensive operations as far as possible (e.g. "2e-5" instead of "2*10^-5")? You want "accuracy and speed", but how much accuracy is sufficient? What about using integral?
For a more specific answer, post the relevant part of the code.