Home -> Magazines -> Issues -> Articles in this issue -> View
Roland S750 Sampler | |
Digital Quality — Analogue FeelArticle from Sound On Sound, October 1991 |
Roland's acclaimed S770 sampler has been quietly growing in popularity, and the release of version 2.0 software and the floppy drive-only S750 should ensure that Roland samplers become still more popular. Wilf Smarties explains why.
The basic machine sports 2 Meg of onboard RAM, enough for around 20 seconds of mono sampling at 44.1 or 48kHz, but you can easily upgrade it to 18 megabytes by purchasing an RAS memory expansion board from Roland (RRP £58), and 4x4 Meg Apple SIMMs from your local chip shop. For those of you who have shelled out for Atari SIMMs the good news is that Apple SIMMs costs substantially less per megabyte (around £45 or so), making the expansion of your internal memory a very cost-effective exercise.
Besides that required to mind-expand your S750, some funds should be set aside for a couple of other peripherals which are really essential to get the best out of the unit. The Roland manual advises you to purchase a hard drive, and so do I. These have fairly clattered down in price this year, and a good 40 Meg unit shouldn't set you back more than £300. Better still, get an optical drive, but wait 'til prices drop next year. Removable cartridge drives are cost-effective but a question mark hangs over their reliability, although retailers assure me that the problems with these devices have now been ironed out. Whatever drive you get, make sure that it is SCSI compatible. Your dealer should know.
The benefits of using a hard drive in terms of vastly increased speed, and in being able to 'nest' files, are enormous, and no serious samplist should persevere with floppies; floppy disks are OK for backing up or importing new library files, but they're too slow for general use. Plus the S750 goes through them like crazy unless you are prepared to buy the more expensive high density variety.
It was always a good idea to go for a monitor with the S770, and with version 2.0 software some of the advanced sample editing functions of the S750 are simply not accessible without one. If you are a tight-fisted Scot like myself you might pick up a cheap S/H green screen (but probably not for as little as the £12.50 I paid for my antique Zenith Data Systems job). A medium resolution colour monitor (c. £240?) gives great results when connected to the digital RGB (SCART) output of the sampler, on which the graphics look not dissimilar to those found in Digidesign's Sound Tools. I read in the manual that the Edit Sample 2 graphics are now hi-res but, unfortunately, I just don't seem to have the spare 1.5K to buy a suitable monitor (and neither will you) so we'll have to take Roland's word for it. There's also a phono jack 'video' output for cheapskates like me. (NB: some domestic TVs have SCART: check it out).
Apart from the exclusion of the internal 40 Meg hard drive, (and a minor improvement in RAM capacity — the S770 can 'only' go to 16 Meg), the only other aspect which sets the S750 apart from the S770 is the omission of the digital I/O facility. This was never as much use as it should have been anyway, because for some reason (assuredly not a good one) it was not possible to back up through this output to DAT. On behalf of all frustrated S770 users I appeal to Roland to amend this oversight. The implementation of SCMS copycode (grrrr!) on domestic DAT machines further devalues digital I/Os anyway. Fortunately the analogue inputs on the S750 are superb.
The S770 has been adjudged to have a sound quality superior to that from CDs. I wouldn't argue with that, and the S750 is no different. It sports 24-voice polyphony (what a difference those extra eight make), and each voice has its own TVF (filter), TVA (amplifier) and LFO. Unsurprisingly these days, it is multi-timbral on all 16 MIDI channels. Sound is managed through a hierarchical system similar to that used on many recent Roland products (terms like TVF and TVA are becoming as familiar now as ADSR used to be). Phase is absolutely accurate between Sample Left and Right, and between the (up to four) Samples that can be assigned to a Partial. (Funnily enough, it's not a problem in mono either).
The level of sound organisation closest to the input jacks is called a Sample. A Partial can contain up to four mono (or two stereo) Samples. Any Sample can be assigned to any Partial. A Patch is a key group of one or more Partials. Performances contain one or more Patches. Now let's look at this in more detail:
The Sample is the level at which sound is brought into the machine. Select mono or stereo, enter a name, set the record level (using the on-screen meters), and go. Options include pre-recording, threshold or manual triggering, and you can stop recording with a simple mouse click. Samples can be truncated, normalised, looped, converted from stereo to mono, and edited in a host of interesting ways which I will describe later. A Sample will play for two octaves upwards of its original pitch, and as far down as your keyboard is long. Thus a sample taken at C5 will cover a span from C0 to C7. In the original review of the S770 (SOS July '90) much was made of Roland's Differential Interpolation Sampling system versus Akai's Variable Rate Sampling. I'm not going to re-iterate the technicalities here, which were well documented in that article, but suffice it to say that transposing a sample downwards on a Roland machine gives a less grainy result, thereby possibly extending the number of usable octaves available to it.
The S750 is a sampler, but it is also a synthesizer, and at Partial level you can impose analogue-type synthesizer envelopes on to raw samples. Here, single-cycle looped waves can metamorphose into bells, strings, percussive noises, filter sweeps etc. More on this later.
Partials (containing samples) are mapped onto the keyboard to create a Patch. A Performance can be used to assemble and mix the (up to 32) Patches employed in a particular song, or it can be used simply as a library file.
The new version 2.0 operating software, which will be a pleasant surprise to anyone struggling with v1.0, uses three levels of organisation: System; Performance; and Sound.
Choosing System takes you to a 5-way menu that deals with such things as SCSI, disk management, MIDI etc. Choosing Performance takes you to a 4-way menu. From here, you can select Play and then choose any of 32 Patches (from four pages of eight), pan them, and assign outputs, levels and MIDI Channels. You can also, via the Patch Map page, access most Patch parameters including (for example) Patch Priority, Filter Cutoff Offset, Tuning, Controller functions, and the intriguingly named Analogue Feel. In almost every case it is the Patch itself you edit, and if you want to keep the edit and the original you'll have to rename it before you save. If you come out of Play and dive into Edit Performance, then open up the COM. window, you can quickly re-ID all Patches/Partials etc. selected via Set Volume ID. (A Volume represents the entire contents of the S750's memory; ID is a 3-character code which precedes the name of the Volume/Performance/Patch/Partial/Sample in question.)
If you are working on a song, and want to keep any edits without losing the originals (which are presumably still on your disk drive) it's a good idea to give everything (except possibly the Samples which are probably unchanged) a new ID before you save. In this way a song's 'Performance' files will occupy only a small amount of memory on disk, as no new samples will have been created.
Even with 24-voice polyphony (and especially if you are stacking samples) there will come a time when there are simply not enough notes to go round. A new feature, Resampling 2, addressed via the COM. window on the Play page, allows you to 'bounce down' everything that you play on the keyboard. By way of an example, consider that mega-snare that you've constructed using several stereo and mono samples panned all over the place. Every time you hit D2 on MIDI channel 10 the snare will sound, but with 16 voices being gobbled up it isn't practical to run it live in the mix. Hit Resample 2, choose Stereo and INTERNAL (source), enter the name "MEGA SNARE 1" and play D2. Now that same mega snare need only occupy a pair of voices. Clever, innit? There is more to explore in the Performance functions, but space precludes an explanation, so we'll skip on to the Sound pages.
On V1.0 software sound editing was done under four headings: Edit Patch; Edit Partial; Edit Sample; and Resample. In V2.0 a fifth page, Edit Sample 2, opens up a whole new block of flats.
There are three pages in Edit Patch. Common and Control offer the same parameters as did Patch Map, together with extensive controller mapping. The Split page has a picture of a keyboard upon which you can map Partials by mouse or MIDI. Options here include Poly, Mono or Exc. 1-16. Mono actually means "polyphonic, but hitting the same key twice will cancel the first note" — useful for keeping note stealing down to a minimum. Exc. 1-16 are 16 mono channels, handy for hi-hats etc. No legato though, so no MiniMoog-type bass swoops (any chance of that changing on a future update?).
At the bottom of most pages you will find five rectangles corresponding to the five Function keys on the front panel of the S750. A left mouse click on any of these selects the indicated page. A right mouse click offers 10 more destination pages which are user programmable, in other words this constitutes a 10-way mark/jump function. Very handy, and very fast.
Wherever you are on the S750, you will find a small pull-down window called COM, to which I've already referred. The COM window is different depending on where you are in the machine, but tends to give short-cut access to things like Disk, Copy, Delete and Edit. Copy and Delete refer to the level you are on, Disk is obvious, and Edit generally refers to one level below wherever you are, so from Edit Patch you can burrow down directly to Edit Partial. When you do this, playing a key on your MIDI controller will call up the partial assigned to that key.
Hitting Exit takes you back to Edit Patch. From the same COM window you can also call up a Partial Map (analogous to calling up the Patch Map from Performance — see above).
There are five pages waiting to be explored here. The first two allow samples to be stacked (up to four at a time), tuned, panned etc. The last three pages are called TVF, TVA, and LFO. For the one person out there who doesn't yet know, TVF stands for Time Variant Filter, A stands for Amplifier, and LFO has something to do with Paul Hardcastle. The first two of these pages look very similar, allowing graphic click-and-drag envelope editing. The TVF filter can be high-pass, low-pass or band-pass. Via the ubiquitous COM window you can call down one of a range of factory envelope Templates to speed things up, and you can also store ten of your liveliest edits as User Templates for later recall. (If only this were possible in Edit Patch with respect to tuning maps.)
There are too many functions in most edit windows to describe them all, but one that deserves special mention is the inclusion of Resonance on the filters. Really what we have here is a synthesizer with decent Roland filters. Sampling, for example, a TB303, then applying some heavy band-pass filtering and finally mapping the TVF cutoff to the pitch (or mod) wheel of your keyboard lets you emulate (and record as sequencer data) playing around with a Bassline in real time.
The TVA envelope needs no special mention, and suffice to say that the LFO can modulate Pitch, Amplitude and TVF depth. (The latter is handy for disguising those single-cycle tail-end-of-the-sample piano loops).
Entering Edit Partial from Edit Patch has one major advantage over entering it from the Sound menu; you can apply edits to all Partials within a Patch simultaneously. You can therefore edit a multi-sampled Patch with a few simple mouse strokes. Think of it as making up a new sound on an analogue synth.
Five Pages are offered, all of them pretty much self-explanatory, namely: Sampling; Looping; Truncating; Smoothing; and Normalising. Rather than bore you with the rudiments of sampling I will allude to a few of the more interesting features. Tip 1: when sampling you might as well be generous with the sample time, especially if you are not sure what's coming, since sampling can be foreshortened with a mouse click. Tip 2: when (forward) looping, after setting up a loop with the mouse switch on Length Lock. You can then scrub the loop around the sample. Tip 3: when you are about to embark on a serious sampling session, enter Edit Partial from Sound, pull down the COM window and select 'Sampling'. Now every time you create a new sample, a Partial will surely follow. This feature maybe doesn't sound as though it has the star quality of Resonance but it is in fact a mega time-saver. 5,000 samples later this is the one I'd want to be stuck on a desert island with. Tip 4: read tip 3 again.
Sampling, as with most functions on this machine, is well thought out, combining ease of use, accuracy and operational speed. Various modes are offered; most include a pre-record function. Both sustain and release loops are allowed, with all the usual features, including autolooping, but smoothing is only available in one direction at present. Any undesirable edits can be reversed via Recover. The much maligned footswitch option can be used to start sampling, and in Previous mode can terminate a continual sampling process.
The Edit Sample 2 menu offers seven options: Loop 2; Patchwork; Digital Filter, Comp/Expand; Time Stretch; Rate Convertor; Wave Draw. Most of these pages utilise a large waveform display where the sample can be scrubbed with the mouse (sounds pretty unhygienic) or made to repeat endlessly in either direction and at unreasonable speeds by pointing with the mouse at a control bar. If you've seen Digidesign's Sound Tools you won't be in entirely unfamiliar territory (even the colour scheme is similar).
Briefly, Loop 2 is a flashier 3-page blowup of Loop 1, which takes its cue from the obviously not inimitable Fairlight III. The full sample can be seen, together with a blow-up of the bit you're interested in. Patchwork allows for cut and paste editing. The digital filter allows, well, digital filtering. The graphic for the dynamics processor (Comp/Expand) is a hoot. No, I'm not going to tell you: go see for yourself. So that's what compression looks like! Timestretch is pretty good and surprisingly fast. Skillfully dummying the boring one (Rate Convertor), let's pass swiftly on to the fitting finale: Wave Draw. Yes, you can now mouse-out those annoying vinyl clicks and pops from your fave break beats, or even put some into your programmed loops to get the punters searching for that elusive source.
Resampling is actually much more interesting than the name suggests. Here you can combine two samples in a number of ways (six to be exact). There are six algorithms reminiscent of FM synthesis, and linear and cross-modulated combinations are possible. Just the thing to experiment with on a rainy Sunday.
Well I think Roland have pretty much had it all their own way up 'til now, so its time to get out the dirty laundry. As an S770 owner, I used to have a much longer hit list than the following, but V2.0 software answered most of my prayers. Even so, there are a few niggles.
• S770 users have long noted a tendency towards 'drop out', or non-playing of the odd sample, usually the left side of a stereo pair. Software version 2.02 seemed to have reduced but not eliminated the problem, at least with regard to the S770. However, in the couple of days I've had to play around with version 2.03, this particular bug has not shown up, so maybe Roland have indeed cured it.
• Floppy disk management is primitive: you can't 'top up' a disk. Saving anything on floppy means having to erase all previous data. The answer? Get a hard drive.
• File management is better in V2.0, but not amazing, and Roland are currently embarking on an overhaul of this area. (Didn't the S770 review say that?) Rumour has it that someone has been looking at how New England Digital do it!
• Patch/Partial capacity could do with being increased all round: in fact I recommend that Roland look into the possibility of providing an equal number of Patches, Partials and Samples. And there is a bug: when Patch capacity is exceeded the last one to be loaded/saved becomes a rogue and crops up in unrelated Performances. Deleting the rogue Patch seems to cure the problem, but it shouldn't be happening in the first place.
• I've said it before, but let's have DAT backup for S770 owners: the hardware is there.
Those of you who have been brought up on D50s and M1s may be alarmed at the lack of digital time domain FX. I for one am not. I always liked the Proteus because it didn't rely on reverb to produce a saleable noise. The MiniMoog wasn't exactly famous for its digital FX either. Tip: use mono outs 5 and 6 as aux sends to reverb and chorus: any sound in the stereo mix can then access either (but not both) FX, with some control over the amount applied. Perhaps Roland could offer true aux send architecture as an option in the next software update? Incidentally, there are plenty of empty IC sockets on a fully-loaded S770/750's RAS board. Could these perhaps be for future DSP chips? I think we should be told.
Like EMU, Fairlight and Synclavier before them, Roland have realised that a strong library is essential to support a quality sampler. Thus they are assembling a diverse and expanding catalogue, ranging from a 16 megabyte stereo piano to a collection of 100 MiniMoogs on one floppy disk. Some of you may have read about the RSS 3-D sound processor. Well, you can expect an out-of-body sonic experience from the S750 soon, as there are plans to create an RSS-encoded sound library.
Roland have adopted a Synclavier-type approach to sampler design, with many of the S750's pages echoing that of the NED monster. The monitor/mouse/menu approach has enabled Roland to offer a working environment second to none that, for the home musician and non-recession-proof professional, must surely represent the way forward. Dealers I have spoken to have indicated that they expect this product to become the machine, and with the first four shipments already pre-sold they may be right.
Comparisons will inevitably be drawn with Akai's S1100, with which the S750 shares many features. The table on the left shows how they measure up.
Whereas Synclavier chose to mix sampling with FM synthesis (and possible resynthesis), Roland have rather merged sampling with analogue-style synthesis. This is indeed a happy marriage, and I am sure that before long noises will be emanating from S750s in garrets that could not have been made on any other machine. Professionals might be advised to look out for an S770 in order to get digital I/Os, but multiple-machine users should only fork out for a single S770; the S750 can take care of the extra polyphony or output requirements, whilst dealing with the 'master' S770 via SCSI. A word of warning: if you can't afford to buy one then don't look at it too closely, or you may end up suffering from terminal envy.
During a period of unusually aggressive discounting earlier this year, I bought an S770. I had a bunch of unwieldy antiques (MiniMoog, Juno 106, TR808, TR909 etc.) and this sampler was the only one that could fool yours truly into believing that he was listening to the real thing. The S750's sound quality is identical — it sounds totally transparent, and I couldn't give its sonic excellence a higher compliment. In common with all the S770 owners I know (and there are quite a few of us up here in the grim north) I could not envisage working with any other system in the sub-telephone number price bracket. Once bitten... Just one question. Whatever happened to the 404?
FURTHER INFORMATION
Roland S750 £2,750 inc VAT.
Roland UK, (Contact Details).
Input Levels: | +4 dBm ~ -50 dBm (variable) | ||
Input Impedance: | 10 kOhms | ||
Output Levels: | Stereo Out: | +7 dBm (1 voice) | |
+19 dBm (16 voices) | |||
Individual Out: | +1 dBm (1 voice) | ||
+13 dBm (16 voices) | |||
Output Impedance: | 1.6 kOhms | ||
Maximum Polyphony: | 24 Voices | ||
Sampling Frequency: | 48kHz, 44.1kHz, 24kHz, 22.05kHz | ||
Convertors: | A-to-D: 16-bit linear | ||
D-to-A: 20-bit linear | |||
Frequency Response: | 20Hz-20kHz (+0,-3dB) | ||
Residual Noise: | Stereo out: | -80dB (all volumes max.) (IHF-A type) | |
Individual out: | -85dB (all volumes max.) | ||
Dynamic Range: | > 87dB (1 voice, rated output) | ||
THD: | 0.01% (A/D/A) | ||
Sound Memory: | 2Mbyte (expandable to 18Mbyte) | ||
Disk Drive: | 3.5" Floppy (2HD/2DD) | ||
Display: | 64 x 20 dot LCD, 200 x 320 dot RGB CRT out | ||
Dimensions: | 430 x 132 x 420mm |
S750 | S1100 | |
---|---|---|
No. of voices: | 24 | 16 |
Sampling: | DIS* | VRS* |
Maximum Memory: | 18 Mbytes | 32 Mbytes |
Mouse: | YES | NO |
Time Domain DSP: | NO | YES |
Resonant Filters: | YES | NO |
Wavedrawing: | YES | NO |
Balanced Inputs: | NO | YES |
Digital I/O: | NO | YES |
Hard Disk Recording: | NO | Possible Future Option |
SMPTE Cue List: | NO | YES |
RRP (inc. VAT): | £2,750 | £3,499 |
* see text.
Hands On: Roland S750
(SOS Sep 92)
Roland S750 Sampler - Stereo Digital Sampler
(MT Dec 91)
Browse category: Sampler > Roland
Roland S770 (Part 1)
(MT Jun 90)
Roland S770 (Part 2)
(MT Jul 90)
Roland S770 Sampler
(SOS Jul 90)
Browse category: Sampler > Roland
Review by Wilf Smarties
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!
New issues that have been donated or scanned for us this month.
All donations and support are gratefully appreciated - thank you.
Do you have any of these magazine issues?
If so, and you can donate, lend or scan them to help complete our archive, please get in touch via the Contribute page - thanks!