Are they samples you have recorded yourself, or pre-existing samples from Akai or elsewhere?
Have you trimmed off all the silence from the end of your samples?
You 'should' be able to play back 32 voices at once.
A mono sample takes 1 voice. A stereo sample takes 2 voices, 1 for left and 1 for right. So if using all stereo samples, you could only playback 16 samples at once.
If you have silence at the end of your samples, it will still take up a voice/voices to playback this silence after the actual sample audio. This silence will waste your available polyphony. So, if you do not fully trim all the silence off the end of your samples, you will be wasting voices, and will run into sample cut off problems sooner than you should.