WS 3.19 (upgradable) WS EX .... 3.19 WS A/D ... 1.25 WS SR .... 1.15You must contact Korg or an authorized service center to obtain these as upgrades.
The problem is with the interpolator overshooting when the amplitude is within a few points of full scale and the slope changes sign. This happens with high frequency sounds and with some loops. To fix the problem, the data is renormalized to stay below the problem region. The gain change is under a fraction of a dB, but must be done at the factory: i.e. you can't just adjust the osc volume to fix it.
The wavestation SR does indeed read the 01 and 03 PCM cards. I've tried both, and it turns out that the PCM's used for the 01 and 03 cards are exactly the same. There is an unfortunate catch; The cards are usually divided into "multisounds" and "drumsounds." The multisounds will read just fine, but the drumsounds cannot be read by the wavestation. Apparantly the SR-exclusive card sets circumvent this by putting the drum and percussion samples in with the multisounds. As far as I know, Korg made six sets of these cards:
The PCM playback chip used in the Wavestation and 01/W series treated PCM slightly different than its predecessor (used in the M1 and T-series). The interpolation technique requires 2 samples *after* the end of the loop, which the M1's PCM chip did not - so M1 cards don't have those extra samples.
When an M1 PCM card is inserted into the Wavestation or Wavestation A/D, the loops are adjusted forward by 2 samples. For some sounds, this will work; for others, it will not. It depends upon the particular sample data being played (specifically, how closely the two samples at the end of the loop match the two just before the loop start). Because of this, Korg does not officially recommend the use of M1 cards with the Wavestation and Wavestation A/D (if you have an M1 or T-series card lying around, though, you might try it out - as above, some samples on a given card will sound OK, some will not).
For hardware reasons, it was apparently impossible to provide a complete fix (I believe that the
main CPU does not have access to the actual sample data, for instance, so it couldn't mock up an
extra two samples....).