Btw this is the error when I load the GUS soundbank conversion. Before the error occurs it lists and loads all the samples up to the mentioned number.

So it is missing sample # 452 but when I open the WFB soundbank with AWAVE it reports only 321 samples and I can't figure out which sample is missing. That's strange.

Perhaps I've an idea where the "too many samples" issue comes from. Back in the dinosaur age when I was converting patches for the Gravis Ultrasound, I noticed that an SF2 instrument can have layers with multiple waveforms. These waveforms aren't exclusive to the instrument. This means 002 Piano can point to the same waveform as 001 Piano. Two instruments, one shared waveform. But the Gravis PAT format doesn't work that way, so on exporting it just makes an copy of the waveform for each instrument that uses it. I think the same thing is going on with WFB. Hence the size increase when going from SF2 to WFB. Shared waveforms/samples are copied, increasing the size and going beyond the sample limit.

Edit: In the screenshot I see you use firmware 1.14. I believe 1.15 is the latest available. Probably doesn't make a difference but still worth trying.

MAUI firmware v.1.15 is attached.

Fabian, you had a broken WFB file (6 300 000 bytes instead of 6 481 276) as stamasd already mentioned. Did you fix it?

Let me recheck again.
//edit: No, the file I tried loading to the MAUI is the same as it should be. 6.481.276 Bytes big. Although the WFSET detects a different size for some reason (The size is diplayed in the picture above, maybe the difference is a overhead of information and SETWF just reports the actual sample size in total). I will try with the newer firmware (thanks!) and report back.
//edit2: With firmware 1.15 the error remains the same (with the same report).
//edit3: I will test again with my recently new aquired TropPlus. Result (Troplus firmware 1.20: Exactly the same!)

Never had anything ICS-ish, but just coughed out the $100.24 for a new-old-stock Maui. Seems I will participate somehow.

Meanwhile, Fabian, what is the date and size of your WFB loadup utility (WFSET or what was its name)? We might be able to find a newer one if any.

Can't wait for your participation!
Well I use the same version as I uploaded in Posting #1, I just checked, it is version 1.43.

Note that Awave Studio is powerful enough to apply any effects (reverb inclusive) to any bank.

I have 82 SF2s under 9 megabytes of size. I took them and batch converted to WFB with Awave Studio. More than half refused to convert complaining that there are too many samples per bank to fulfill WaveFront standards. Nevertheless, here are remaining 36 WFB banks: http://geryon.ru/wfb_8mb.zip.

Whoever wants them soundbanks converted, send them my way (err, too many Witcher 3 lately).

It seems that there is an alternative (unofficial) piece of software to manipulate the Maui (should also work with Tropez, dunno about Tropez+), including WFB load. It's called SETMAUI and has an additional graphical interface to it named MAUICTRL. Find them attached.

SETMAUI readme excerpt:

1USING SETMAUI
2-------------
3
4  SETMAUI [ [/|- options]  [filename [,x [,y [,z ]]] ]  [/|- options] ]...
5
6  []  means optional
7  |   means exclusive or
8  ... means etc.
9
10  If you don't add any parameters you'll get a short message on how to get
11    help.
12
13  If you specify a filename it means you want to download a file.
14
15  Now the valid extensions are:
16                                     .WFB      - Bank file
17                                     .WFP      - Program file
18                                     .WFD      - DrumKit file
19
20  - .WFB files
21
22         The software download all samples, patches, drumkits and programs
23         that are in the Bank.
24         This action may overwrite the internal samples.  See NOTES.
25         You need not specify any other parameters because all rooms for
26         the samples and patches are fixed in the .WFB file.
27
28         Sample: SETMAUI gm_maui.wfb
29
30...

Thanks for your work, bristlehog!
I tested the FANTA 8GM.WFB on the Troplus with both SETWF and SETMAUI and both result in the same error as described before, but this time it stops loading at sample # 425. Very strange!
To rule out the possibility that anything is running out the cards sample RAM I tried the converted Chaos 2MB soundbank which resulted in the same error, this time at sample # 144. The Anson Lam's 512KB soundbank fails at sample 101. Arrrgh 🙁

See the below picture of SETMAUI:

I also switched between 24 and 32 voices with no difference. At one point the Tools stop loading the soundbank.
Maybe its worth mentioning that any soundbank reports 128Prgs/1 Drumkits/256 Patches and 512 Samples). Maybe, just maybe the WFB format expects 512 Samples to be present.

I'm wondering if the RAM you're using with your card may be faulty. Can you try another pair of sticks?

bristlehog wrote:

Never had anything ICS-ish, but just coughed out the $100.24 for a new-old-stock Maui

Wow that's a lot of money. Sorry for that. I managed to get the last card still priced at $22 from the same seller- seems that just after I bought that he bumped the price to $31, and then to $45.

I think the sticks are ok but I will try other sticks.

//edit: Now testing with 2x 1MB sticks instead of the 2 x 4MB sticks I had in the cards before. Result: Unfortunatlely nothing changes. The small banks still refuse to load, the error remains the same.
Another observation: Loading WFB files requries Extended Memory which is allocated before the samples begin to load.

There is another method to upload the samples to the cards RAM: Via the MPU-401 interface. This means very slow transfer, like over the waveblaster header with the RIO board. Maybe this works.. Now trying to load the 512KB bank, having 30% within 5 minutes - woohoo 😁

stamasd wrote:

I'm wondering if the RAM you're using with your card may be faulty. Can you try another pair of sticks?

It might be worth a check, but I suspect that Fabian's RAM is ok. What's supposedly not ok is Awave's work with WFB (who will ever believe that any bank has constant number of samples!) But it's still but a guess.

I applied a quick and dirty fix to that GUS WFB bank, give it a try: http://geryon.ru/gusdirty.zip

stamasd wrote:

Wow that's a lot of money. Sorry for that. I managed to get the last card still priced at $22 from the same seller- seems that just after I bought that he bumped the price to $31, and then to $45.

It's still $45 but I had to pay $55 for shipping to Russia.

Thanks! I will. just waiting for the result of the 512KB bank.. still loading.. 😉

//edit: 512KB bank still failed after loading all the samples.. still couldn't find sample #101.
//edit2: Well bristlehog: YES MAN! Thanks a lot! Loading completed! Midis soundend all right in GSPLAY! (did skip through a few standard midis. descent sounds odd but I remember it sounding off also with the GUS.. The whole bank sounds like a GUS with effects processor! Of course quite similar to the original patches of the Tropez+, but that was expected. I think that was the first huge milestone in our project - finally a working converted soundbank!

Screen attached below - This time it reported 451 samples and not 512. What exactly did you do and can it be done in a batch to all of your .SF2s?

FGB wrote:

Screen attached below - This time it reported 451 samples and not 512. What exactly did you do and can it be done in a batch to all of your .SF2s?

I just changed the number of samples in WFB header to a number of samples actually present there (451, judging by the fact that sample #452 failed to load).

Number of samples is a word value located at 48h offset. So, I changed 200h (512) to 1C3h (451) and here we are. I don't yet think this can be applied in a batch since I can't figure how to determine actual number of samples in a WFB bank apart from trying to load it and receive an error at (n+1)th sample. However, we might tackle it later. For now, we can [probably] quickfix any Awave-made WFB bank by putting the right number of samples into WFB header manually instead of omnipresent 512.

I'm not even sure that such a fix won't break anything. We've got to experiment more.

OK I understand. In the trial version of AWAVE one can count the actual samples with "view ---> item count". GUSDIRTY.WFB reports 129 instruments and 321 waveforms. Thats 450 in total. I think that SETWF and SETMAUI just count both together as SAMPLES. So these together are 450 and the drumkit make it 451: Et Voila! Thats the same number you edited.
So keeping this in mind makes it now very simple to find out the correct number for the 48th offset.
I will "dirty edit" the FANTA 8MB soundbank and / or the CHAOS 2MB soundbank. The numbers for the offset are 424 and 143 .. I think we get closer. If I'm right both banks will load and work! Can't wait 😁

FGB wrote:

In the trial version of AWAVE one can count the actual samples with "view ---> item count". GUSDIRTY.WFB reports 129 instruments and 321 waveforms. Thats 450 in total. I think that SETWF and SETMAUI just count both together as SAMPLES. So these together are 450 and the drumkit make it 451: Et Voila! Thats the same number you edited. So keeping this in mind makes it now very simple to find out the correct number for the 48th offset.

That was my first guess too, but nope, it won't work this way. The Fanta GM WFB contains 65 instruments and 234 waveforms, but 424 samples according to your screenshot, while 234+65+1=400.

It would have been so easy, too easy as it seems.. 😒 I edited the FANTA 8MB bank to 424 samples and it loads but some instruments are missing and making strange noises, like the Synth Strings. Other instruments play well. Still a lot of work to do I guess..

I can think of two guesses right now:

1) Awave Studio won't write correct samples number to a WFB bank, always writing 512. (But Fanta GM won't probably glitch then).

2) It's required to always have 512 samples in a WFB bank. Awave Studio just won't complete the bank sample structure, writing only samples that are physically present. But then default gm_maui.wfb, which is not an output of Awave Studio, contains 332 samples, not 512. (There's also "332 of 512 samples in use" string in SETUPSND output at your screenshots).

I'm bewildered.