still confused on stacking sounds in program mode

[kanthos]

Not in the real sense of it, maybe it does a better job then the others but not in the real sense cuz thats phisically impossible, same as you moving from one room to another and being able to be in both at the same time.

I haven't tried the Fantom G myself; I only said it has patch remain based on how Roland themselves marketed it.

And no, it's not physically impossible; I have no idea what you're talking about there.

A normal keyboard flow is something like this: keys -> oscillators -> envelopes -> effects -> line output, right? (I realize I'm simplifying; that's not important). Here's one way of doing patch remain: duplicate everything other than the keys and line output so we have two complete and distinct signal chains. Call them A and B. Add something that sends key events to only one signal chain; call this the Splitter (for lack of a better name). And add the circuitry to merge the sounds from signal chains A and B at equal volumes before sending them to the output. The Splitter works as follows: it knows what the current signal chain is and knows which signal chain each key last sent to. Any further events on that key up to and including a Note Off are sent to the signal chain previously used for that key; any Note On events are sent to the current signal chain.

When I play the flute, say it's going through signal chain A. Say I'm playing a C4 and that the flute patch uses aftertouch for vibrato. When I press C4, the Note On goes to the splitter, which redirects the event to signal chain A and marks C4 as going to signal chain A. Aftertouch continues to be routed to signal chain A.

Now suppose I change to my electric guitar patch without releasing C4. The splitter marks signal chain B as current. Aftertouch messages for C4 still get sent to signal chain A, even though it is not the current signal chain. Now suppose I play G5. The splitter sends the note on to signal chain B, since G5 was not currently pressed and B is current. When I release C4, the flute stops sounding, since the note off goes to A. If I then play C4 again, the note on goes to signal chain B and the splitter marks C4 as directing to signal chain B in case I switch patches again.



It's a fairly simple scheme; the drawback is the cost of having two signal chains. My suspicion is that any keyboard that has patch remain will start with something like this and look for ways to combine components rather than have to duplicate everything, which makes the keyboard cost more.

Still going to say that it's not physically possible?

[BasariStudios]

I understand what you are saying, yes it is possible but on the machines
there is still basic things which are impossible, forget that. Your scheme
is simple to understand but not easy for this machines to do it. They freeze
on very basic operations not on something complicated as that.

[X-Trade]

Not in the real sense of it, maybe it does a better job then the others but not in the real sense cuz thats phisically impossible, same as you moving from one room to another and being able to be in both at the same time.

I haven't tried the Fantom G myself; I only said it has patch remain based on how Roland themselves marketed it.

And no, it's not physically impossible; I have no idea what you're talking about there.

A normal keyboard flow is something like this: keys -> oscillators -> envelopes -> effects -> line output, right? (I realize I'm simplifying; that's not important). Here's one way of doing patch remain: duplicate everything other than the keys and line output so we have two complete and distinct signal chains. Call them A and B. Add something that sends key events to only one signal chain; call this the Splitter (for lack of a better name). And add the circuitry to merge the sounds from signal chains A and B at equal volumes before sending them to the output. The Splitter works as follows: it knows what the current signal chain is and knows which signal chain each key last sent to. Any further events on that key up to and including a Note Off are sent to the signal chain previously used for that key; any Note On events are sent to the current signal chain.

When I play the flute, say it's going through signal chain A. Say I'm playing a C4 and that the flute patch uses aftertouch for vibrato. When I press C4, the Note On goes to the splitter, which redirects the event to signal chain A and marks C4 as going to signal chain A. Aftertouch continues to be routed to signal chain A.

Now suppose I change to my electric guitar patch without releasing C4. The splitter marks signal chain B as current. Aftertouch messages for C4 still get sent to signal chain A, even though it is not the current signal chain. Now suppose I play G5. The splitter sends the note on to signal chain B, since G5 was not currently pressed and B is current. When I release C4, the flute stops sounding, since the note off goes to A. If I then play C4 again, the note on goes to signal chain B and the splitter marks C4 as directing to signal chain B in case I switch patches again.



It's a fairly simple scheme; the drawback is the cost of having two signal chains. My suspicion is that any keyboard that has patch remain will start with something like this and look for ways to combine components rather than have to duplicate everything, which makes the keyboard cost more.

Still going to say that it's not physically possible?

it is a nice repsesentation on patch remain but I think most synths (for example the way the triton series *attempts* to do this) actually use polyphony. for example notice that on the triton adjusting the envelope has no effect on notes that are already sounding. but then other multitimbral synths work a bit differently too, but it is based on each voice having some fixed parameters that are coppied across (e.g. the envelope on triton) to that voice's parameters, and the rest is just a pointer to the timbre that you are using for it.
so basically when a key is hit, it loads into the next free poly note some data including a marker that says which timbre this is playing from.

the reason it still doesn't work in most keyboards is the effects - because they are not polyphonic!! you would need either a polyphonic effects section which would require extreme processing power and be expensive, and would probably have less insert effects overall, or you need two effect chains as you suggested. it is the effects chains that is really the problem...

[Cutter]

ok so then what happens when you switch sounds on the Roland while holding down the damper pedal?

[X-Trade]

reading the link, it says that each timbre in their 'live mode' has one effect (there are eight timbres), so i guess it does have something to do with their effects system, although i've never touched the board let alone had time to try to figure it out in action.