If it was a synthesiser you would think it was a stuck key. Continuous
cyclic racket, sounding like a swarm of hornets, mosquitos and bagpipes
with chorus effect thrown in. Presumably DSP problem. Changing buttons
etc changes the character of the noise and the display functions
normally. Only got in far enough to see a RED LED flashing fast and
regular but slightly modulated I think, and noise does not vary with
twizzling.
I'll take a recording of it before full disassemly, in case it is
diagnostic

The effect is on harmonize and double functions only , recorded for
reference. No different whether ground lift button is activated or not
and whether a proper ground connection is made and the LED flashes all
the time. It is full level intrusion , not background level.

When the distortion is there it emerges after POST or not present after
POST and stays in that state till switch off. When present there is a
phasing stereo image of it on the phones output and adding smoothing to
the DC input or adding a true grounding makes no difference.
Of course no way of forcing the outcome one way or another at switch on,
because that would indicate where the problem was.
Sometimes R ch is just crackle in sympathy with the full noise on the L ch.
When not present and adding a badly screened mic input and the gain
turned up then a similar sort of noise gets in as background hash.
Time to take the unit apart fully

same old digital rigmarole. No scrubbed idents and datasheets for
everything probably, including the DSP. Other than close inspection of
soldering, checking 1V and 3.3V supplies and grounds and any
poer-on-reset devices not much else other than random insulated SMD
squashing and lead tugging etc at power-on, thats all that can sensibly
be done. I will try tracing the didtortion back to a possible origin,
assuming with the boards opened out probable, the distortion returns.

Perhaps a reset problem. Touching header J301 p10 with a probe, kills
the noise and normal function seems to resume.
The LD1 red LED flashing seems to be normal function.
AT POST the sequence is 2 second off, 1.5 sec flashing, 1.5 sec off,
then continuous flashing.
Finger injection of EMI the noise goes immediately and flashing LED
goes off for 2 sec, 1.5 sec on, then continous flashing.
p10 connects round the houses of the DSP board via a jumpered J101 for
some other option, to what I assume is some sort of control processor
with 32M of memory, has Helicon label v1.2.01 on it.
Under is Spansion S29AL032 then looks like 070TF103.
Datasheet not found on spansion.com but their numbering seems to be
letter not 0 (or O) after the 032, but suffix TF103 part is similar to
other devices of theirs.
Will trace the other way to the display board which is still in the casing

seems to be
s29al032D70TF103

Assuming pinning of S29AL016M is the same, then pin12 is a reset.
Removing the 2 top boards, the system micro is a
ATMEGA 88 20AU
on the underside of the board. The "EMI reset" goes to p32 of that,
which is an input also, so perhaps a flakey long period reset IC/
electrolytic somewhere. The 3 electros on the top board ESR'd ok and not
connected to that line anyway