when i do a long record multi part of the performance Anubis on highway in the recorder [ 15 to 25 minute ?] when i stop the recorder,
montage freeze,no alert message, i waited few minutes but nothing happened { happened twice ]. , had to reboot. is montage recorder overloaded by this patch ?
dont know. here the patch ANUBIS on highway i use.

I assume since you are recording for 15 to 25 minutes you are using the Audio portion of the Play/Rec Performance Recorder.

I mean attempting to record 15-25 minutes of MIDI just doesn't sound wise. Let us know. Sorry, I've not heard your sound, is there something specific about it that would cause your issue?

Recording Audio gives you approx. 74 minutes record time (depending on your USB stick).
Recording MIDI gives you approx. 130,000 notes (which you are eating up with controller movements, no doubt). I don't know how many controller event equals a note, sorry, but I'm sure it's a finite number which you apparently have exceeded.

If you are recording MIDI, try coming up for air every 5 minutes or so, instead of playing until even the sequencer stops paying attention 🙂
Even recording audio to USB drives can be affected by whether or not your USB stick has a large contiguous area into which you can record.
If you wish to record that length of time, (15-25 minutes), and you must do MIDI, I recommend recording to a DAW on your computer - then you have no worries about a limit of notes/events. Make sense?

thanks for answering. it is about midi record. this patch have many real time controllers. if it is a memory amount limit there is no alert message, just freezing? i hope it is only a memory issue but we should have some alert message and know what are the limitation too. and it should not bug. some work for montage os programmers ? or maybe memory hardware;) defaults ? for sure i will use daw for long record. this recorder is quite anecdotal compare to the huge potential of montage .may be i understand now why there is not a proper sequencer in the montage, it will have need stronger processors for handle the information flux, isn't it ? 😉

Sorry, I've not had the time to push the Montage in this area, I frankly, never would record that much MIDI data, 15-25 minutes... it's not like I'm gonna go through it and edit it or print it out as notation. I can't even say what is normal when you just push it it like that.

Please record the same thing (or similar) to a DAW and see if you can determine the .mid file size it creates.

Perhaps a loud gong sound to let you know you're out of memory, didn't one product announce , "Bummer, dude" when you ran it out of memory?

It should reprimand you or at least announce the error, so we'll see what they say. (Love to include approximately the file size though, if you can, thanks).

Number of "objects" recorded is a good theory. There's also a framework around this which may change depending on tempo even with no huge note traffic. Right now, I'm letting the recorder run. I played some notes here and there, but it's mostly just recording basic timestamp information without any real bus activity. It shows over 26 minutes and no lockups. I did record some knob movement on a PWM modulation part I was working on in another thread.

What's your tempo for this recording? Mine is 90bpm. If yours is higher - than something that's more tied to the position # as a problem would depend on the tempo. It may very well be you're running into the max events to record - but wanted to be sure I had as much as your setup the same. The sound itself you made public - so that's easy to load up. My testing was starting with a random performance because I'm not sure yet there's anything specific in your performance that would "crash" the recorder vs. any other performance. But who knows.

Any arps applied, motion sequence, super knob automation? This can eat up events.

BM - a great use case for the recorder would be when learning a set of user performances on the band-stand. You may want to record your tunes so you can playback and look for scene changes, etc. I know the last gig I botched my scene changes because I didn't write the new Montage control presses on my charts (like press "scene 7 here"). So I may want to look back at what my intuition did and resolve if I made the scene changes too difficult to remember and maybe edit my scene mapping. Or look at other parts I played for whatever reason.

The recorder is a fairly blunt tool since it locks out so much of the interface. But at least if a single performance is used for a song (no performance change) - then perhaps ok for this type of recording.

No tune is going to last that long (well, we do some 4 tune medleys so these can get long) - but the last tune of the gig or of the set - maybe you forget to turn it off. Depending on if events, time, or measure count is the culprit here - this may or may not make a difference for running into the issue.

I will go in kicking and screaming before carrying a laptop or PC into a gig. There's nothing great to record scene changes and the like unless I have a camera on the keyboard that can capture all this. Not exactly the best use of technology there either since I really just care about the MIDI - not even the context of the other audio (band) for this application.

I think if phones could integrate with Montage - I'd be OK with using that. Or if a tablet could be used as a touchscreen - I'd go paperless for my charts (touchscreen) and integrate a tablet into the live environment. There's a lot of synergy around tablet + keyboard if Yamaha could work this out.

I know the original thread is the recorder - but using a PC was brought up as a possible workaround - so I wanted to dig into this a bit more and offer my use-case and perspective.

Assuming he did exceed 130,000 so-called notes limitation, it shouldn't cause the Montage to freeze. If this is what is happening, it will be easy to reproduce with any performance if, that is, one wishes to simply exceed the seq midi events limit. Just pick his performance or another performance and just play until the seq reaches its limits. Has the Montage team not done this? They should have. No way a seq should lock-up. Motif xf doesn't.
Remember that the keyboard action when struck in a certain way, can cause pressure events to be recorded and these type events use up memory, The FSX type keyboard is more sensitive to generating un intended pressure midi events than most other keyboards that I've used.

The most likely scenario is buffer overrun which has gone unchecked. The easiest test for this is to setup some automation (arp + superknob automation) since that would automate lots of messages. Unlike recording audio (which records to external memory - a USB flash drive) - MIDI recording uses an internal buffer which is where the limitation comes in. You get to record as much as was malloc'd, or if dynamically allocated, until you run out of the wider pool of memory Montage is reserving each chunk within. Allowing to go external for both would be good policy - at least to have a choice for MIDI (internal memory vs. stick if detected).

Also kind of interesting is what I assume to be the elapsed time display (to the right of the buffer position graph). This has lots of error for me. If the tempo is slow, it advances much slower than real-time. At bpm=120, it seems to match real-time. At bpm=300, the clock runs very fast. I would think this elapsed time indicator should scale itself so it always is showing Minutes:Seconds rather than Seconds*Tempo*K/60:Seconds*Tempo*K

Where K<1 to scale the effect of tempo. Maybe it's not "supposed" to be MM:SS - but that would certainly be strange.

I'm currently running a test where I turned ARP HOLD = ON for "DJ Montage", turned knob automation ON, and set amplitude = 127 and also the pulse curve scaling factor to 127 for the pulse so superknob runs (ramps) from 0-127 and repeats.

I wasn't "watching it" - but somewhere around position 1000, a few things happened:

1) position reset back to 0 (and is now counting up again from 0)
2) The buffer position graph stopped moving
3) At position 148 (after the position reset occurred - and thus this is an odometer "wrapping around") the recording stopped. It stopped my ARP - so all is silent now. Superknob automation is still running. Note: technically, this is position 1,148.

Note: this is not so much a position "reset" as it is that the display only shows 3 characters for position although internally the variable can be above 999 - so I "see" it go to 000, but the 1 in the thousands place (for 1000-1148) is not shown. I doubt here there is any overflow since 1,000 is not a power of two.

At 1148, with this setup, I reach some kind of boundary condition.

Montage did not lockup (although the recorder stopped and notes stopped automatically once the position counter reached the end of 1,148). It fairly gracefully stopped. I could go back and play what was recorded - all the way to 1,148. Upon playback, although the position was reset to 001 after completing playing back position 1,148 - the ARP did not stop. The sound continued to hold the ARP - which was slightly different than what happened when recording. Presumably when recording something "bad" happened so Montage may have sent a MIDI reset which stopped the notes. During playback, I suppose I did not "go over the edge" so there was no reset and the ARP continued to play.

I did set the tempo - sometime in the middle of the recording - to 300bpm so I could get more notes to get recorded and not have to wait so long. I would suspect that a less busy recording would allow for a longer recording (past position 1,148). 1,148 does not strike me as a natural boundary although if I consider my meter (4/4) this is 1148*4 = 0x11F0 which does start to perhaps look like it could be a boundary condition. Only real way to test is to try turning ARP off all but one PART and redo the test. If it's a position-based boundary then I'd fail at the same 1,148. If it's event count based, I'll get much further before the recorder decides to quit.

Now I've turned ARP from "hold" to "sync-off" for most PARTs except for one synth and drums. There's a lot less going on - although still enough to have some activity. It takes a while to get to position 1,148 - so I'll leave it here for now.

Certainly features should have 100% test coverage - but "unfortunately" people still make and test these devices. Therefore, you do have cases where bugs get out.

Keep in mind my firmware version may not match - so a later version may have introduced the non-graceful ending once the buffer (or position) ran past its end.

BTW: you should be able to turn aftertouch off to prevent those unwanted events from clogging up MIDI / the recorder.

@Jason: as an aside, this whole "I want to give in-depth technical advice but your results may vary because I still haven't updated yet" is actually pretty amusing at this point.

This is the fifth or so time you raise this caveat. What's your aversion to updating? You certainly have enough time to post and experiment - but more important however is that your advice that you put so much effort into might not apply to the latest os.

Update: the test has been running with less events happening (due to some ARPs turned off) the position is now at 1,330 and still running (elapsed time, which is off due to above discussion, shows around 107:00 ). Certainly it has not been 107 minutes since the last post.

So it seems like, at least here, only strange things with the recorder happen when an amount of events is exceeded. Even then, I cannot reproduce an unrecoverable lockup - but someone else with v1.20 can do the same test. I would suggest turning on all ARPs (arp hold) and setup the superknob automation so there's lots of movement. Then you can just let it run without touching the keyboard.

(Position 1,577 now with 125:00 "elapsed time" - in quotes because this is not true - clock is running fast.) ... still going.

Michel - I understand your amusement. I place the caveat in order to be most informative as it would be reckless to provide feedback without the reminder. Yamaha or another forum member can decide on their own if my observations are valid or not. In most cases - the help I give is generic - but when it comes to corner case type stuff - certainly firmware version can come into play.

The minutes it takes to perform a firmware upgrade is not the factor which leaves me at an earlier revision.

You could see the glass as half full as well. Having previous firmware allows feedback to Yamaha on how something may have broken with newer firmware relative to older firmware if the behavior is not the same. This would be similar to regression testing. One of a handful of motivating factors is that Yamaha has indicated that I cannot return to this firmware level for regression testing. So until either: a feature I need cannot be supported by this firmware, or the downgrade-to-this-fw-level barrier is lifted - I keep the reluctance to upgrade. Thanks for your understanding.

My comments about how to easily generate a test that anyone can perform on their own is firmware agnostic.

... an update here: I did see that the record button is "locked out" after recording a long sequence. Changing performances doesn't bring back the ability to record. The entire interface isn't locked up - just the ability to press record (can still press play assuming the record buffer has something in it).

Oh my!

So, the seq clock does not count or show properly at the measure depending on what tempo is set and it varies? hmm. Well, please do tell is this true? The on-board recorder is a big disappointment and I wish to state that I do not care that Yamaha says, this is what it is. Yamaha MAY change their position at any time and I will continue to ask for some expansion here. Sure, it make sense that if one has 8 parts, one can record all 8 straight-away. The XF did the same from its performance as well, you could record all 4 parts BUT, for the Montage one should not have to always record the same number of tracks as is the performance- switching on/off RECORD status per track if needed. We should be able to find maybe a single part performance and place that on an empty track, etc...very very basic stuff here folks. What are the other 8 tracks for? If its just a music performance instrument, then why even have a recorder? If you have a recorder then don't go out of your way to dumb it down to the point of questionable use. Just bring it up even a little, which would really help. To say that it is just a recorder is not even true since you have replace, overdub and punch in/out so, why not have Yamaha run a Poll or other survey regarding seq

You can overdub the other 8 tracks (9-16) one at a time.

The inability to wipe clean one track - or inability to use the "Play [ON]" buttons to dual as a "Record [OFF]" feature - then use "record type: replace" (which would only replace the "Play [ON]" tracks) is not so optimal - I can see that. The board, for me, is too heavy to carry around in the bus and use whenever I have inspiration. The gig isn't where I find inspiration - at least not to do track recording either. So this, for me, leaves the studio where recording would happen and I'm possibly in the minority who is OK with using bundled software for a sequencer or making an additional purchase for some other software solution.

I do know that Yamaha is listening and maybe some incremental changes can be made which help improve the user experience. In the meantime, it would also be great to "shore up" the functionality that exists today and improve function reliability (assuming there is an issue).