It is not just that audio quality will not change, neither will the quality of the recorded softsynths. An increase in the sampling rate from less than 44.1 KHz might arguably increase the quality of audio data from any source. I say arguably because the point is frequently argued here as to whether a further increase in the sampling rate above CD quality will yield more accuracy in reproduction. If you softsynth has the kind of antialiasing filters on its output that most AD converters have on their input so that the synth output is for all practical purposes a band limited audio range signal, increasing the sampling frequency above the Nyquist probably does not improve accuracy.
In any case increasing the bit depth of the sample will never improve accuracy, but will only provide a greater dynamic range. Since the dynamic range of the 16 bit audio representation is pretty much everything you can hope for in terms of sounds from just perceptible to literally ear splitting, 24 bit does not improve the representation of music, but it gives an enormous cushion against clipping.