Hi leeperry, I need to make a correction.
The ppm specification is temperature stability. The frequency accuracy is not really specified.
However one can assume that for a crystal specified at 25.576 MHz, any variation in the last 3 significant digits is permitted. Say 25.576499 MHz still qualifies as a 25.576 MHz part. Thus 500 Hz of variation constitutes 20 ppm.
On top of that you add the frequency variation from the value at measured temperature to the value at operating temperature. Lets say it was measures at 25C and you run your DAC at 45C. That could be another 10-20 ppm for a 100ppm part
And then there is aging: 3 ppm the first year.
The 22.5792 part is better because the maximum variation from spec is 50 Hz which is only 2 ppm
Of course this is just theory. There maybe other contributing factors beyond the crystal/clock