A method of operation used by some digital synthesizers that employ digital signal processor integrated circuits, in which the DSP resources are dynamically reassigned according to the complexity of the currently selected patch and the chosen effects. Doing so allows the synth to increase polyphony when a less complex patch is chosen. As an example, if a given synth has 16 DSPs and all of the processing required for one voice in the current patch can be done by one DSP, then the available polyphony is 16 notes with that patch. But if a more complex patch is chosen that requires three DSPs to process one voice, then polyphony is reduced to five notes (with one DSP unused).
Some performers find DSP flex to be very frustrating, since the way that DSPs will be allocated is not always apparent to the performer, and on current synths that use DSP flex, the polyphony available for a given patch can be found only by actually trying the patch. As of 2013, no current DSP-flex synth has any means of indicting to the performer how many voices will be available while the performer is building the patch.