You have to understand I'm not familiar with this device at all, the only Roland box I have tangled with was a MKS7 which is a very different thing.
I see that Q17 is not a regulator as such, it is a capacitance multiplier (the “ripple filter” in the block diagram), and there is going to be some drop across that, so when they say “9V” for the output there I think they may be exaggerating a bit!
Somehow the processor seems to be alive, but not quite. If you can find why there is nothing on the LCD I think you will have made a big step forward. I see there is even an LCD test routine - if only you could get something on the LCD in the first place.
At first I thought the LCD was one of those very common HD44780/KS0076/etc modules. Well, it looks very similar, but the connections seem to be a bit different. Another reason why you might only get blocks on the display could be that the LCD operating voltage is too far out. Often there is a “contrast” preset to adjust this, but here the voltage, V5 or LCDV5, appears to be under control of the processor via PWM. So maybe check the LCDPWM signal from the processor and what is going on around Q5 to LCDV5.
I still don't know whether there is a definite problem with IC6. The LSD0-LSD7 lines are also involved in scanning the switches, and I thought perhaps you could be seeing signals there relating to that rather than driving the LEDs, However, looking at it again, maybe those lines are only used as inputs to the processor for scanning the switches. The real indicator is still whether you see any of those lines high when you see the clock for IC6, but the corresponding output stays low, and I'm not sure from what you said if you actually saw that.
Another clue that the processor is alive would be if you see anything happening on the outputs of IC5 (scanning switch rows).