Roland FA-06/FA-07/FA-08

Well, besides all of the issues described in INTEGRA-7 page, there are some new thing, that got notably fixed, but far from completely or even really usable.

It is kind-of supposed to be an evolution & revolution over Fantom-X and Fantom-G, and fixed misdesigned issues of Integra-7, but it's really not yet there at all. So many years of Roland R&D should have yielded some dazzlingly awesome stuff. But it did not.

Some things mentioned below were fixed for the FA, but just in a half-assed way, so realy need (bug)fixing.

Most importantly:
‣ Old Fantom-X effects were ported to it to one additional insert fx block (IFX), and that is superb, but it misses the control assignment page. Meaning, currently piano sympathetic resonance pedal control cannot be assigned to a pedal... Problem is, the piano desperately needs it. Even the SuperNeutered one could have been made somewhat better with it.
‣ Annoying thing is that the most useless - SuperNeutered patches are always first, not User ones. This is just patch dialing&sorting issue, and should be solved.
‣ Also a deep design flaw is that for some dim witted reason, User patches are separate for normal (good) extended-JV architecture and superneutered ones as separate banks... That could have been easily avoided, since all superneutered (including synth) do have less parameters and therefore data amount than the classic patch, and so fir in the fantom patch data block footprint perfectly well. Why would it matter to divide those - user patches should just be a monolithic block where each can be whatever it needs to be - "PCM", "SNA", "SNS" or whatever.

‣ Don't expect SuperNeutered patches to be any useful in anything serious - they are not. There is no editing, no real parameters or options, besides stupid few preset variations (for no reason, since mostly those are way too different to even be a in a single patch, with nothing in common like legato, pizzicato and staccato strings as variations in a single patch) and a macro parameter at best. SuperNeutered Synth is just an extremely limited subset of JV voice features, besides adding supersaw (oh, glory...). To be useful, all "acoustic" SuperNeutered patches would need their parameter pages not hidden, so stuff can get tailored to at least something useful. Currently, you cannot tailor anything at all, so those cannot be used in any production - only to play at weddings & funerals.

‣ Regarding the point above, I can tell you (and I do know everything in it in tiny details) that ARX superneutered patches and "technology" is entirely the same old SRX. It uses everything the same, same samples (a lot from old collections, and better if so, since new sampled ones are not realy good, or even low-grade licenced with huge watermarking holes as the horrid "environmental effects" in INTEGRA-7), occasionally abysmally processed (improper use of FFT-iFFT denoise to split sample sets into main and residual parts) and all that there is about the "technology" is that actual JV patch parameters get adjusted through an extremely limited preset macro-parameter (or a few - like for drums), and sometimes there is some awkward software addon that switches over samples or enables portamento and stuff based on playing - which, in fact, is that "super modeling" the ads are referring to. ARX was such an absolute fail even in its idea, and then even worse - implementation, that there's not much that can be fixed about it. In synths that has that, those things should get moved far away even after GM patch banks, and waveforms (all ordinary, there is no modeling or different synthesis involved whatsoever!) should be selectable for the normal extended-JV architecture patches (intentionally named as "PCM" to make you think superneutered ones are not) so you can make smoething useful out of them and that's it.

On a positive note: I still have it, and there are definitely some good aspects to this. As a sound module for custom patches with classic-jv-based stuff, it can be used.
‣ IFX shoud definitely get (bug)fixed, though.
‣ Keybed is horribly godawful.
‣ Encosure is more flimsy, bendy and quickly loosing its looks (gloss scratches even from fingers and fingernails, rubberized parts deteriorate etc) than you'd expect from Roland.
‣ It is, just like Integra-7, based on the dreaded Roland SSC chip (essentially Super Sound Canvas, developed for use in home organs!), which don't realy sound that good - nowhere near Roland WX chips from Fantom-X and all the high-end goodies from those years. Haven't yet pinpointed exactly what's up with it, but there has clearly been some "optimisations" in its design that affected something. So don't expect it sound anywhere near anything WX-based. It kinda sounds flat and lacking any splendor and liveliness.
‣ Loading old SRX boards is just for the waveforms (which is still great) but patches cannot sound the same (because of completely diffferent fx stuff, and not only that) so if you are after some exact SRX patch sounds, just get an appropriate old box that can fit a real one or something.
‣ Again, there's something wrong with how SSC renders the waves. Don't expect the effervescence, sizzle and sparks of the XV and WX sound, despite seemingly better specs.

Stuff below this line is still unchanged from the INTEGRA-7 page, but most of the stuff still relates to the FA as well. I'm going to edit it someday probably.

Fantom XR has excellent sampling system integration, allowing to create user multisample sets and use them along rom waveforms. Integra-7 - FAIL
Since JV, XV and Fantoms, there should be much more interesting MFX algorhithms, right? Wrong. Reusing all old stuff plus removed some - EPIC FAIL
Fantom XR can hold 6x SRX plus half a gigabyte user waveform data. Integra-7 can only hold 4x SRX and zero bytes user waveform data - FAIL
Surprisingly, most roland synthesizers miss Vocoder effect. Roland has has some of the best nailed down on DSP as early as 1993. Still not there - FAIL
Fantom-X and many other WX-chip based synthesizers had a very good multiband mastering section. Integra-7 has only master EQ. - FAIL
Sympathetic resonance was a new good stuff in Fantoms for a realistic piano sustain simulation. Can you find it in Integra-7? No. - EPIC FAIL
On all older synths, that had more than one MFX block, you could cascade them. Integra-7 has a plenty of MFX blocks yet no cascading. - FAIL
While SN Synth is kind-of supposed to be equivalent to, it does not hold par with the old JP-8000 (introduced in 1996) or even simple XV structure - FAIL
XV-5050, Fantom X, XR and G had a poor visual appeal with all that silver, plastic and wild western styled parts. Integra-7 looks great. - WIN
On older gear, preview phrases were fixed for respective categories. For some weird custom sounds you sometimes need them differ. Integra-7 - WIN
Polyphony is the same. Still the same like in XV-5080 or even EMU Proteus 2000. It's not bad, though - NEUTRAL

There is not only much room for improvement, but also a lot has to be done to even just be on par with Fantom XR.
The added Motional Surround is a good thing, but not extremely useful. Those virtual 3D things (virtual speaker, phones) are not usable for real productions, while 5.1 channel music is not realy a necessity. In films, center channel should never be used for music, and therefore two Individual / Aux outputs on any usual sound module is enough for adding special rear sounds if ever needed.

"All SRX boards" being not all at all.

Good that it has most of the SRX boards available for loading. But how about SRX-99, SRX-98 and SRX-97?
Yes, those were kind-of-demo kind-of-promotion boards that in most cases had only a few megabytes of waves for some extra fun. But why not to include them, especially if all of them together would take less space than a single normal SRX card?

SRX-98 "Analog Essentials" manual states that all waveforms have been newly sampled (including some 92-note multisamples), making it a member of the SRX collection. Not including it clearly brakes the "all included" claim.

I personaly would be happy to see all SO-PCM and SN-U110 card waves available as well. Many of them are classic after all. And I don't mean patches or oldschool "timbres" - just waveforms available when editing a usual JV/XV/Fantom type 4-partial sound. SN-U110 cards do take only 512kilobytes in their original 8bit logarhythmic form, and when converted to uncompressed 16bit would take only 15Mbytes for all 15 cards. SO-PCM uses floating point style compression, packing about 24bits into approximately 8.1bit. There were about eight SO-PCM cards, each being 1 or 2 Mbytes compressed. So here we get about 20Mbytes uncompressed. Add some classic synthesizer ROM waves, say D-50 (512kbyte), D-110 (1Mbyte with all MT-32 waves and more), U-220 (3Mbytes) and MC-303 (why not?), maybe even throw in all SR-JV80-xx boards in unaltered forms and this sound module will have a dazzling appeal for any synth zealot.

Slots for SRX

Having an option to load four additional SRX boards is nice. But why only four and not eight? Is DRAM so expensive these days that 256Mbytes even matter? Address space of the synth chip wavebus is not filled up as well, so no excuse there either.

The six slots in Fantom XR were appealing for a reason. For example, for some soundtrack work you might want to fit Orch, World, Brass, a pair of electronic music related ones as well as an additional piano or strings board. Along with Fantom XR waves and editability this is more than enough, even not counting the massive sample loading capabilities. Four slots instantly shows up to be a pain.

Secondly, the whole point of ROM wave expansion boards is to have all of them instantly available, simultaneously playable and browsable through at any time. While load time of about 4-5 seconds per one SRX is quite ok, a need to plan and choose only four of them or less is an obstacle to creative flow.

Synth for synthesis

Why do I call the SN Synth a SuperNeutered one? Because it's essentialy what it is.

First look gives us the (marketing targeted) impression of it being somehow a JP-8000 or a classic box equvalent. That it might somehow do great stuff like KORG R-3 or Alesis ION/Micron does, for example. No, it can not. Not even come close.

Oscillator SYNC? No.

Is there a Feedback Oscillator like in JP-8000? No.

Are there any means of True waveshaping or Driving (like in a prehistoric KORG 01R/W and all their later analog modeling ones) for oscillators? Even though simple waveshaping, quantizing or decimation doesn't realy eat up any DSP resources - No.

Seriously? No SYNC and no nonlinear driving? Roland, is it the best you can do?
Preset "sync" waveforms does not count, because they are not controllable and cannot behave in any interesting way. Waveforms with sync are as much a substitute for real hard sync, as there were waveforms (in old simple romplers) subsituting for a lack of filter.

Ring modulators are present and available in all usual four-partial voices (since JV), so there's nothing special. Additionally JV, XV and Fantom classic voices have FXM cross-modulators and simple driving-waveshaping distortion (Booster) blocks that are not present in SuperNeutered Synth.

How come a specific super-über-natural "Synth" mode in almost every aspect be inferior to the most usual structure since JV-1080?
Oh glory, it has a supersaw.

More MFX blocks = good. But is there flexibility and variety?

Unfortunately - no there isn't much. For some reason, even Sympathetic Resonance is gone. It seems to pop up here and there among SuperNeutered sound parameters, but I don't realy care about them. No vocoders (see below), no PSOLA-based harmonizer/pitchshifter for dramatic electronica stuff. Almost all effects are mundane and usual, since they are from XV-5050 times.

There are some electric piano combination effects added, but it's just a recycling with small alterations.

Some glitching and stuttering effects (sometimes called "grain shifters" or whatever) that essentialy are a short or medium length delays with a different control would have been a good thing to make the sound module more appealing in todays dubstepped glitchcored market. Even though such processing takes about a tenth of resources required by a simplest chorus - No.

You can make some grainy effects when cascading two or more pitchshifters with extreme settings. Oh well - you can't cascade MFX in Integra-7. Facepalm.

Essentialy anything besides channel/part insert fx looks like Sound Canvas 88. And not even 88Pro, as there are no additional AUX or Insert effects. SC-88 actually had a bit more than Integra-7 - it had an aux Delay processor besides Reverb, Chorus and EQ. Yet another facepalm.

Vocoder is still missing

In my opinion, Roland has best vocoder DSP algorhithms - ones that combine outstanding intelligibility, transient response and no modulator leakage (leakage is a problem present in all FFT vocoders as soon as they are set to high band counts necessary for them to sound with any intelligibility). They have significantly better intelligibility than reorganized simple synth banpass filter banks used in all KORG-s like MS-2000, Radias, microkorg, kaoss stuff or R-3.

Vocoder is an interesting tool for a synthesizer to have. Impressive and immersive pad textures, synthesizer sounds and dirty electro madness can be achieved by using it. Even old synth workstation KORG Triton, introduced in 1999, had vocoding available for it's insert effects. So it shouldn't be considered something unusual or weird. It actually can be considered as a kind of necessity.

Roland had nailed down a very good vocoding DSP algorhithm in Boss SE-70 back in 1993. The old JP-8080 has one. In 2005, the VP-550 had awesome sounding one. So where's the problem? It have been successfuly made/ported on most of their synth chip architectures, including their major chips like ESP and WX (these sound engines behind variphrase, VP, JP, VSynth, Fantom e.t.c). Code is done and working - why not to use it?

PSOLA based pitchshifter/harmonizer is missing as well

Pitch Synchronous OverLap-Add algorhithm based pitchshifters and harmonizers maintain formants in place, or allow shifting formants without altering pitch. These effects sound great for electronic music and unusual soundscapes.

Again, Roland already has it available since mid and late nineties - remember VT-1, VF-1 effects processors and VS8F expansion cards? DSP code is done and working, and could easily be ported to any of their newer chips. So what's the fking problem?

Imagine a formant shift as an insert effect, steered via matrix modulation - for one component of a complex pad/motion sound... Or a rhythmic pulsating stuff that alters formants for some partial(s)... Or a drum kit. PSOLA is also useful for surreal violin, woodwind and brass sounds. A lot of Variphrase-feeling effects can be obained with this one.

Multiband Mastering block is gone

It is a realy good tool that allows instant finishing and polishing of whatever you are creating.

Outside of the part instert fx blocks, Integra-7 realy looks like Sound Canvas 88. Not even SC88Pro. I mean - entire processing section is just Reverb, Chorus and EQ..?

User program/patches

There is some confused forumer (or just a troll) with a nick EvilDragon, that posted in several forums a nonsensical claim: "And what's even dumber - only 59 user slots". Nobody even wondered what the f does a user slot mean anyways. People are so easily goalable (trollable).
[Shot1] [Shot2] [Shot3 "Still not stupid enough for you to believe"] etc

However, as I started to check around for user patch, program and perfromance amount, I was surprised to see that it is never ever mentioned anywhere... including Integra-7 manual and specifications. Not even where user patch saving is describled, and in no screenshots do you get an idea of a range. This gave me a strange feeling that there actually could be something terribly wrong with it. The only excuse I could think of might be that it was not yet known at the time of writing of a manual and web release stuff.

Everything got cleared up by taking a close look into midi specifications book for it, where exact numbers should allways be found. There is a plenty of user patches and performances after all:

USER PERFORMANCES (called studio sets) - 64
USER SN ACOUSTIC PATCHES (super unNatural acoustic) - 256
USER SN SYNTH PATCHES (super Neutered synth) - 512
USER SN DRUM KITS (super reNamed drums) - 64

As you can see, there is a plenty of space. Though not more than in previous modules, like Fantom XR. Actually much less because Fantom X did directly address patches and kits off PCMCIA cards with 2000 or maybe even 5000 spaces available.

Patch mess

As you might have noticed above, there are way too many different major categories for patches. This might as well be a reason for omitting these numbers from publishing in web and manual specifiaction - it's a mess. And those numbers are not realy ones that somebody in year 201x could be proud of.

All fine regarding distinction between a drum kit and tone/patch... Oh and preset and user can normaly reside in different banks as well. Thats ok.

But why would an end user care for (pseudo-) technology behind the patch? If I want some of my patches to be stored sequentially in some order, without ridiculous categorizing and planned naming-for-the-purpose, I just need to be able to store anything anywhere within user patches. Be it a SuperNeutered synth or JV-structure based (Four partials with boosters, filters, ringmods and FXMs) one - I need them to be right beside each other.

Only after entering a patch edit do I care about what is behind it.

As supernatural patches are by design so limited and barely costumizable, byte amount it takes to store one is less than for a classic XV/Fantom voice. So there was/is no problem to just reserve a fixed space in every user patch, that can be of any "technology" behind it (the largest being XV/FantomX patches and kits anyways). A main "mode" parameter for every user patch could select everything that (and how) is editable about it... and that's it. Simple and smooth.

Demo songs - some XV-5050 orchestra, Cheesy Castrated Electronica

Are all japanese patch and track designers gone from Roland? Some indeed might be, as you can see by recent investor relations reports (available on Roland main and japanese websites). But there must still be somebody that can drive this box into more interesting fields than that. It actually feels that most songs and patches are designed by one guy, who, while being a good auto-accompaniment style and pattern composer, might be somewhat too old and soft to produce real appealing stuff.

The power and strength is missing. There is no bite, no fangs, no brutality or hell unleashed. Back in the days, japanese guys could rock even a basic sound canvas based brick like mc-303. Or a Sound Canvas 88's DSP based SH-32 - remember it's demo "Spank"?

While this does not directly affect it's use as a tool for composing, such demos show for what patches, effect system and concepts were designed and targeted for.
Elevator music? Low to medium budged TV ads? Radio friendly Chese
-Hop, Faggot-Pop?

The orchestral demo "Film Scape" does not differ from XV-5050 orchestral demo (I think it was called "Turbulent") to quite a scary extent.
Jazz stuff is good, though.

Oh, but there are so many PWM and pseudo-Sync leads and pads! Sigh and a pacepalm.

Or maybe it indeed is intended to be the new Sound Canvas?