The MONTAGE is capable of reading Voice/Performance, Waveform and Arpeggio data from a Motif XF All Data File (.X3A)
The Motif XF data can be loaded either to the MONTAGE User Bank (where you can work on it) or to one of eight Libraries.

The User Bank has the following limits (if you exceed these, your data will not load as saved).
When set to “Import Option” = Voice, you would opt to convert “Voice” so that you have searchable access to each of your 512 XF Voices.

All 512 User Voices in the XF File are converted to 512 Single Part Performances in the MONTAGE (There is a maximum of 640 Performances in the MONTAGE Bank... that’s five sets of 128).
2048 Waveforms maximum. This could be an issue.
256 Arpeggios. That’s the same as in the XF

When set to “Import Option” = Performance, you could wind up in trouble with the Waveform Limits. I can envision a situation where a Drum Kit Voice with tons of Waveforms and Samples, If duplicated in several Performances that could do it.

If your Motif XF All Data File is a .X3A that would mean your Waveform Limit was 2048 Waveforms per FLASH Board. This means you should save your Motif XF data in two separate files so that you wind up a .n1.X3A and .n2.X3A — so that no one File exceeds the 2048 Waveform Limit. (In the XF you can opt to save an All Data with or without Samples... create two files so that all your Waveforms can be represented between the two)

Sorry I don’t have any experiencing loading in Performances that house custom Waveforms from an XF... I do know when you use Import Option set to Voice, the MONTAGE is able to create a searchable catalog entry for each XF VOICE that contained custom Waveforms. If the Waveform is used in a Part of Performance, the conversion will possibly create multiple versions of the data. I can see that happening.

You don’t provide any specifics about the number of Waveforms or where they are stored...

I only have data stored on one flash card on the Motif and the waveform total is just over 1700, total data in the X3A file is 182 mb. When auditioning the waveforms on the flash card in the Motif i did notice that there were duplicates of certain drum sounds. If this is the reason, this should have been tested long ago, so users would not have to waste days trying to figure this shit out. There is zero data being used in my Montage internal memory-it shows 0.0 out of 1.8 GB. This is a mess. So what's essentially happening is Motif can easily store duplicate files on the flash card, but there is no way to import them into the Montage flash memory. Now I have to go through 1700 + sounds and delete the duplicates on a screen that only shows 20 or waveforms at a time-scrolling one by one, not even knowing if my X3A file will import after I do it. As a Motif expert you should know about possible issues with regards to loading custom waveforms into the Montage from an X3A file-since it is an function of the Motif. Where do I go to find answers if I don't look here? Yamaha seriously dropped the ball here by not instructing you to learn all functions of the Motif-then seeing how the transition to the Montage would go. Did Yamaha just expect users to forget the work they had in progress in the Motif and miraculously re-create it by trial and error on the Montage? This is waste of my time ffs.

You can call Customer Support.
Without seeing your data I can only guess based on the partial information you supply. If it is only 1700 Waveforms it will not exceed the User Bank Limit, if transferred efficiently. What you possibly are missing is the significance of “Import Option” = Voice versus Perf. I’ll try once more, then you go do what you want.

If you translate the MOTIF XF data using “Voice”, the MONTAGE will build a Waveform List (without duplicating any Waveforms). Each XF VOICE will be converted to a MONTAGE Single Part Performance. This is the best (most efficient) way to convert XF data to MONTAGE imho... because each program item becomes searchable individually. When you Import Performance, then the 2nd, 3rd, and 4th Parts are cateloged linked to Part 1.

There is no VOICE mode in the MONTAGE. Therefore, if you only load the data by “Import Option” = Perf, you lose the ability to have a separate *Searchable* item by Voice Name for each program. Sorry, if this is a bit difficult to grasp but by not having a Voice Mode reference system, this impacts the results when you opt to convert just the XF Performances. Each time a program is used in a Performance — it is stored autonomously. This is MONTAGE-way (it has exponentially more memory space)...

I simply do not have any data that only exists in Performances and uses lots of custom Waveforms. Therefore, I can’t tell you first hand what’s down that rabbit hole. I simply converted *my* XF data as “XF Voice” to “MONTAGE Single Part”. This way I have each converted in the most efficient manor to use in the MONTAGE.

I simply do not test all the inefficient ways you can possibly transfer the data. I can only pass on how other methods of converting the data might be the cause of your woes. I really cannot tell you exactly based on your provided information.

If your XF File were made in a Motif XF, it would have an extension .n2.X3A or .n1.X3A if only one Flash Board was used. You said that the extension was .X3A —which if Flash Boards are installed, denotes data on both boards. This is why I suggested you might have exceed the 2048 Waveform limit. I don’t know what happens in that situation because, again, I don’t personally have any data that would exceed the 2048 Waveforms Limit (not even between 2 FL Boards... I was kind of curious what would happen... and no, I don’t test every contingency just to see *what happens*. I do know the 2048 Waveform Limit was per FL Board, and is now the Limit for the User Bank.

If all you have is 1700 then this is definitely not your issue. Import the XF data as “Voice”. This is the most efficient way to translate your XF work to a MONTAGE. Once in the MONTAGE you immediately have so many ways to improve it... that was my personal concentration and direction. I then build my own multi-Part Performances in MONTAGE... it’s a new synth, if I wanted it to be exactly the same as in the XF, why... Well, I’ll just say, that was not my goal. I’m looking to show and feature “what’s new” about the Motion Control Synthesis Engine.

There is no one way to work. And no one person could anticipate or test all the possible scenarios. I personally gravitate to the “most efficient” and then those I try to pass along to the interested. If someone were to ask me how to best convert their XF data to a MONTAGE or MODX, that contained lots of custom Waveforms, I’d recommend setting the “Import Option” = Voice.

All Waveforms will be imported, those in User 1-4 Voice locations in the XF become 512 Single Part programs in MONTAGE. Those User Sample Voices recorded to the XF’s Integrated Sampling Sequencer will show up in the Waveform List unassigned to Performance. You will see that 512/640 Performances are filled, but when you go to the Waveform catelog and set the Bank = your XF Data... you will find all the Waveforms that we’re converted.

If you have further questions you can always call your local Yamaha Customer Support, (that should be your first stop, actually). Then you can open a case and someone will work with you through the transition.

I would hope if Voice with custom/user waveforms called "V-Custom1" is used in 5 Performances that when Montage converts the Performances it only stores one Waveform and each Performance that used "V-Custom1" points to the same single waveform instead of replicating the waveform. It wouldn't make much sense to create duplicates when performing one conversion operation.

If you split your conversion over two steps - Voice and Performance - then I could see the chance for duplicates here. Convert the Voice and 1700 waveforms land (no "duplicates" ) in the sense of any more than the original. Then convert Performance - there's the same 1700 waveforms. I could see this 2nd conversion step not sifting through what's currently installed to find matching waveforms. That's a bit much given what I know about how the system works and is implemented. This second operation shouldn't create any more than 1700 more waveforms. But the total of having both (Perf and Voice) would be 1700x2 = 3400 > 2048. I'd hope Voice "import" (load) would work. I'd also hope that, Performance "import" (load) would work too assuming, for this data set, you didn't already import Voice. The feedback seems to be that Performance loading from your X3A file - even without loading the Voice data first - causes the memory full error. If that's the case, this would appear to indicate something inefficient in the conversion.

Error messaging could always be better and more verbose (for my tastes). This rubs both ways.

Unfortunately, this is part of that observation that traditionally the Yamaha synth file-system has always been one of the least user-friendly pieces of the overall usage puzzle.