I will play backtracks from my Montage, by loading a Stereo-file splitted (L) click and (R) backtracks, and then trigger it by the key E0 and I want it to play until the end of the whole song.
Is there any reason why I shouldn't do this or will it work?

I can think of a few... You don't really explain why you want to accomplish this like this. Did you know...
_ the Montage features Main L&R and Assignable L&R...
_ the Montage click can be routed to the Assignable Output
_ the Montage can generate Tempo from the A/D Input
_ the Montage can stream audio from a USB stick connected to its TO DEVICE port
_ the Montage can stream audio from an iOS device connected to its TO HOST port
_ audio connected to the Montage A/D Input can be processed and routed to its own discreet output (Asgn L&R)

I'm not exactly sure what your ultimate goal is but I would avoid any situation where I had to Pan the click one way and the music backing track the other. I would at least attempt to make use of the dual stereo outputs to maintain stereo.

Anyway, I have imported the waveform of 03:42 seconds and 37.4 MB filesize to my performance. I created a drum-init and set the Decay 2 to "Hold", and Rcv Note Off to "Off", but it stops between 00:05 and 00:07 randomly all the time. I did this in another Performance with another file and it works perfectly. I guess there must be something I have missed because I did the first Performance of a Preset with the Montage, and the second one by editing another Performance.

Could be several things, without a bit more info we would be guessing.... Alternate Group, polyphony, it's hard to say, but you might have inherited something in the Performance... (?)

What is the ultimate goal, there are so many options ... but I guess it will depend on exactly what you need accomplish.

Loading: https://www.ee.columbia.edu/~dpwe/sounds/mr/pzm12.wav into E0 - this seems to work fine - went well above 7 seconds and still listening to the fingernails across the chalkboard that is this wav file. I was looking for something around 3:42 - and the search engine spit back this 5-minute recording.

I set attack time to 0, decay 1 time to 0. Decay1 level to 127. Decay 2 Time to Hold. Receive Note Off as Off. I set assign mode to single just so it wouldn't muddy-mix if I retriggered the same key. Group is OFF because I am not concerned about the turning off the sample by pressing a different key - but you can use group between a couple keys (maybe F0) and assign F0 to any waveform - set the level to 0 - and this will "turn off" the recording when hitting F0.

EDIT: adding - I started with the "Init Drums" part so there was no residual unknowns from editing a performance.

Okay, thanks for your answers!

Bad Mister, I will use the keyboard and play with the backtracks, and I want FOH to be able to mix the backtracks separate from the keyboard, that's why my only choice is to do the BT mono.
L/R internal keyboardsounds, Assign. L click and Assign. R backtracks.
It would be the same if I play it from an iPad or something similiar.

I tried to copy the same performance that works to play the backtrack.
But when I import my second .wav file I still have the same problem. It stops after a few seconds.
The only thing that's different with this waveform is that I haven't put anything yet on the right channel, so it's empty/quiet.
Could that be the problem? I tried to bounce a new version so I don't think the audiofile itself is the problem, it works on the computer.

And it doesnt help if I hold the E0 all the time, it stops anyway. Seems to be randomly, sometimes it almost doesnt start at all, just play 1 second and often the count-in and then stop.

Quiet, is your enemy when playing back samples. All samplers have some kind of silence detection... when the audio signal stops for a certain amount of time, the instrument declares a note-off and returns the resource to the polyphony pool. When the DECAY 2 is set to HOLD... it will hold until it meets a block of silence.

"It works on the computer"... the computer does not play it as a sample. The computer plays the .wav the same every time. A sample is started by a MIDI note-on which determines its volume, 1-127, at the very outset. That note-on also determines its playback pitch, and the duration. The computer treats the .wav as a fixed entity so it is not similar when you really look at it. Also the "silence detection" thing takes affect with samples... HOLD is not truly indefinitely, it's until the audio goes silent. What determines "silent"? A block of no signal... "empty" is not allowed.

Okay, so there is no way to change that "silence detection" in the keyboard?

Just add "silent noise". Something that will keep the silence detector from tripping - but is still not distracting or audible.

I'm sure one can come up with a waveform that would fool the silence detector but would sound, to a human, to be silent.

If you want to directly change how this works, short of moving to Japan and applying for a job on the firmware development team as someone reasonably high up on the chain enough to dictate feature-set, I'm not sure what you can do beyond suggestion either here or through other channels available for feedback to Yamaha.

I wasn't aware this feature was there - but assuming it is - there's no way to turn it off that is user accessible that I can see.

Robin wrote:

Okay, so there is no way to change that "silence detection" in the keyboard?

No, it's not a user programmable function. We have not actually determined that this is the cause... if it were, as you said, "empty/quiet" that would trigger the silence detection, but normal audio clips (samples) do not contain "empty" space, also if it were the cause it would happen precisely in the same location every time, not " it stops between 00:05 and 00:07 randomly all the time"...

So I'm not ready to conclude that this is even the cause of your issue. Let's see if we can nail it down a bit more, then put effort into a workaround.
Can you give us a better idea about what the nature of the audio is in this sample? Does empty mean literally empty?
You don't need to move to Japan or join the staff, just describe the nature of the audio in your clip.

Samples are typically started by the note-on; you are placing audio in a Drum Kit... think about this in terms of how this feature would be optimized to playback Drum/percussion audio. As close to the very first sample (@44,100 per second) the Note-On will start playback. It does not like several seconds of silence out front because to be musically useful you want immediate response. Drum/Percussion Samples are programmed to be triggered on the clock pulse that audio begins... therefore the functions here are optimized for interacting with typical Drum/percussion sounds.

Drums/percussion instrument sounds are typically short - turning the HOLD On could eventually be problematic if there were no Note-Off to end the audio... eventually you will need that note back. The difference or one difference between Normal and Drum Parts is how they can deal with messages from the Keyboard. Drummers look at a whole note on their music chart differently from "Normal" musicians (feed your own drummer joke here), to the normal folks it means to hold that tone for the count of a measure, to a drummer it means don't hit anything else for this duration of time.

So the normal musician can literally 'hold' a note. Drummers don't deal with that function. To feel what I'm talking about take a crash cymbal and set it so it Receives note-off... you'll quickly see how annoying it would be to play most Drum and percussion sounds if they receive note-off. Try a kick and snare, it makes it almost impossible to play them without trouble. We are so very used to Drums completing their audio without us having to hold the keys.

As piano players (we are actually percussionists, as well) we are spoiled by the sophistication of our instrument's ability to break this mold - we are percussionist who deal with duration by the nature of the Key and the mechanisms that allow the piano strings to ring out.
Yamaha also has a special "VibraphonePedalMode" (also caused piano players to scratch their heads) when playing vibraphone the Sustain function works different... hitting and holding notes would mute the sound, hitting and letting go in conjunction with the Sustain pedal would cause the sound to ring... most piano players were baffled by it so it has not found its way into many products: CP300 has this, for example..

That aside, let's see if we can pin down exactly what is causing your sample audio clip to randomly stop... let us know.

Re: VibraphonePedalMode

Are you kidding me? Pleeeeaaaaseeee bring that one back along with new vibes and mbira samples/performances.

not exactly the same thing - and not key-to-key (instead "channel") - but you can setup aftertouch to squelch your AEG decay 2 so "pushing in" to the keys to engage the aftertouch will create the "thud" sound of muting with mallets.

I used the "Vibes" performance and added source=aftertouch, curve = standard w/ratio -64, destination = all elements (on) AEG Decay 2.

A ratio of -13 is maybe more natural.

There's bad side effects of using this approach since it does affect all sounding notes. So if you're holding "ringing out" notes - then mash in for the thud - your ringing out notes also get cut off. But in certain situations - this kind of control may do something you're after.

Could always split and maybe leave the left-side to ring out without this feature and do more single-note in the right hand with the "thud" behavior.

Could probably get fancier and setup a way around this (better mousetrap). Unfortunately, velocity offset can not be a destination or I'd offset the velocity to do some tricks to force one set of elements to play vs. another using aftertouch as the deciding factor. I'll have to think more about if there's still a way to get there. Trying to stay away from XA control since the goal I'm setting is to make this a gesture you can do with the playing hands and not go "off key"(bed).