Magazine Archive

Home -> Magazines -> Issues -> Articles in this issue -> View

Roland S770 Sampler

Article from Sound On Sound, July 1990

Roland's flagship sampling module boasts an impressive specification, and its page-based screens are oddly reminiscent of the famous Fairlight. Kendall Wrightson finds out how impressive it is in the flesh.

Roland's flagship sampler has arrived, a full year after the initial announcement. It boasts an impressive spec, but has it been worth the wait? Kendall Wrightson has the answer.

When Akai launched the S1000 stereo digital sampler back in 1988, the only true rival was Emu's Emulator III, which cost nearly three times as much at £8,000. With no significant competition, the past two years has seen the S1000 become the market leader and an industry standard. However, Akai's machine now faces serious competition in the form of the Roland S770.


Roland actually announced the S770 over a year ago, leading one to assume that the premature announcement was something of a marketing tactic designed to slow sales of the Akai S1000 until the S770 was ready for the shops. If so, Akai's recent announcement of the S1100 (for August delivery) may well hinder Roland's plans as the S1100 matches the S770 spec in many respects - and if you can wait a year, an extra three months is a mere bagatelle.

Nevertheless, the S770 is available now, fully operational in every respect and already receiving its first major software update. Has it been worth the wait? Read on...


In designing the S770, Roland have been careful to appeal to both the musician and the studio engineer. For the latter, the S770 offers 48kHz, 16-bit sampling with 24-bit processing and a 20-bit DAC, a RAM capacity of up to 16 megabytes, and a realtime digital audio output. For the musician, a sampler's sound library is the most important consideration - as the recent success of ROM-based sample players illustrates. By making the S770 compatible with the S550/S330/W30 and offering CD ROM and erasable optical storage from day one, Roland have managed to please the musician too.

A sampler that meets both the engineer's and the musician's approval is therefore guaranteed to appeal to the producer. For all three, the onboard 40Mb hard drive, 24-voice polyphony, and excellent user interface make the S770 easy to understand and lightening quick to use. Then there's the S770's unique facilities, like 'previous' sampling, multi-mode filters, and resampling... but let's start with the basics.


The S770 is a 3U high rack-mountable black box which not only looks cool but also feels cool - literally cool to the touch. This is due to a powerful internal fan which can be set to automatically turn off when in sampling mode. Without its customary 19" rack-mounting 'ears', the S770 looks more like a piece of hi-fi gear, but Roland do supply end cheeks for those who need them.

Figure 1.

The front panel contains a bare minimum of controls because Roland provide fast graphic software which can be displayed on a normal television (in black & white) or in colour on an RGB monitor, via rear panel sockets. The front panel 64x240 dot LCD actually displays part of the 200x320 pixel CRT output, and is therefore continually scrolling, because all the major commands are placed at the very top or very bottom of the screen as you can see from Figure 1, the default Play page (which appears 35 seconds after power-up).


The display is divided into three parts - a main menu at the top of the screen (in reverse video), a window displaying the current page with its own menu, and five small page select buttons at the bottom. To navigate around the display (CRT or LCD), Roland supply a two-button mouse. The function of each button is displayed on the right-hand side of the main menu - 'Inc/Dec' in the case of Figure 1.

To change the value of a parameter (say, the Level of the marimba in Figure 1), you simply highlight the field by positioning the mouse cursor over it, then use the left and right buttons to alter the value. A single click increments/decrements by one unit, while holding down the button causes the values to scroll rapidly.

The left-hand side of the top menu bar always displays Mode, Index, and Mark items which, unlike Inc/Dec, are 'soft' switches - when pressed they reveal smaller sub-menus.


Figure 2.

The first of these switches, Mode, offers three alternatives: 'Performance' for normal playing, 'Sound' for sampling/editing, and 'System' for editing MIDI, SCSI settings etc. As you can see from the Mode Chart in Figure 2, each mode has four or five main functions, and each of these has up to five pages of sub-functions. In addition, each function has a further set of mostly disk-oriented parameters which are accessed by clicking the Com switch in the left-hand side of the page menu bar.

To the right of the Com switch is the Exit switch, which simply closes the current window. The left end of the page menu displays the current mode and page ('Play' and 'Page 1' in Figure 1). The five page select buttons at the bottom of the display access the sub-function pages, scrolling the entire page to the left or right.

With so many pages and windows to remember, Roland offer two shortcut methods. The first is an Index window (a permanent main menu item) which lists key words and functions in alphabetical order, eg. aftertouch, filter, etc. Clicking on a key word reveals a menu of relevant pages, and having made a selection, the S770 jumps straight to the correct page.

The second shortcut is Mark/Jump, which offers the facility to store five frequently used pages. This is done by selecting 'Mark' when in a desired page. From then on, pressing the right mouse button or the front panel Jump button changes the page select switches to display the currently available 'marked' pages.


The large backlit LCD can be navigated without the mouse, using the four front panel cursor buttons. Parameter values can be changed quickly with the rotary alpha dial and/or the S1 and S2 switches (which mimic the left/right mouse buttons). The five soft keys located directly below the LCD duplicate the page selection switches.

Of the remaining front panel buttons, Index, Mark, Command, and Exit operate exactly as they do in the CRT screen display. The Sound Play button plays the currently selected sample in the sample edit pages, and the Graphic button turns the LCD off when not required, which speeds up the general operation of the machine.

Unfortunately, Roland do not supply an RGB monitor in the price, but you'd be mad not to buy one, as it makes the S770 a real pleasure to use. If you can afford to pay £4,800 for an S770, an extra £300 shouldn't break the bank!

The S770 can also be operated using the RC100, a remote control originally designed for Roland's S330 sampler, which offers a numeric keypad and alphanumerals (via a Shift button). Both the RC100 and mouse plug into the front panel Ext Ctrl socket located in the recessed lower section; this also sports an LCD contrast control, stereo headphones and footpedal sockets. The Left and Right ¼" jack sample inputs provide alternatives to the rear panel XLR sockets, and a Sensitivity pot adjusts the sample inputs to match line or mic sources from -50dBm (hard right) to +4dBm (hard left), with LEDs to indicate clipping.

The front panel also houses the 3.5" floppy disk drive, which can read both 2HD (1.6Mb) and 2DD (640K) disks. The S770 operating system software runs from its internal 40Mb hard disk, so the floppy drive will only be used to load in software updates and library sounds. Hard drive access and MIDI receive LEDs complete the front panel features.


The S770's rear panel XLR sample inputs have a three position input gain switch to accommodate +4dBm, -10dBm, and -50dBm operating levels. Surprisingly for such a professional instrument, the six individual polyphonic outputs and the main stereo outputs use ¼" jacks rather than XLR connectors.

The optical/coaxial digital I/O (input/output) makes the S770 directly compatible with DAT recorders and CD players fitted with digital outputs, and Roland's own R880 Digital Reverb and E660 Digital Parametric EQ. However, in the studio (which is the S770's natural home), engineers will also want to transfer audio to and from digital multitrack and mastering machines. With its real-time digital output ability, the S770 could also be connected directly into a digital mixer, so it seems a shame not to offer a wider range of interfaces. At the very least, it ought to have an AES/EBU interface. In fact, the rear panel includes an interesting blanking plate which is large enough to take the requisite couple of XLR connectors. In the meantime, a digital 'standards convertor' such as Audio Digital Technology's FC1 will be necessary.

Just below the digital I/O connectors are the RGB and composite video outputs, plus a 25-way SCSI port. MIDI In, Out, and Thru sockets complete our tour of the rear panel.

Figure 3.


Figure 3 shows the S770 data hierarchy. The lowest form of life is the humble 'Sample' and, regardless of memory size, the S770 can hold a maximum of 512. Also stored with the Sample are two loop points (sustain and release) and the key (note) number at which the Sample was originally taken. The S770 is advertised as 24-voice polyphonic; for voice read Sample.

The next level up is the 'Partial', which is constructed of up to four Samples and roughly corresponds to a Keygroup on the S1000, and a Primary or Secondary Voice on the Emulator III. The Partial allows each Sample to be independently tuned and also stores the TVA, TVF, LFO, and velocity (SMT) parameters.

Up to 88 Partials can be mapped across a keyboard to make up a 'Patch' - roughly equivalent to a Program on an S1000 or a Preset on an EIII. However, S770 Partials cannot overlap like Akai Keygroups or Emu Voices, so positional crossfading and layering must be set up within a Partial.

Each Patch can be assigned its own MIDI channel, and a 'Performance' holds up to 32 Patches to make a multitimbral MIDI setup. A collection of up to 64 Performances is called a 'Volume', ie. the complete contents of the S770 RAM (whether 2Mb or 16Mb). The internal 40Mb hard drive can hold a maximum of 128 Volumes.

Although the internal RAM can hold 512 Samples, these Samples play through Partials and only 255 Partials can be stored in RAM at one time. In a large multitimbral Performance with lots of percussion and multi-sampled Patches, 255 Partials may well prove limiting.

One slight irritation with the S770's hierarchy is that loading in one level of data automatically loads all the levels below it. For example, a Patch will also load all of its Partials and Samples. This makes loading a 'dummy Patch' or 'Patch template' impossible, which is a nuisance since the ability to copy a drum map (Patch), for instance, would make life much easier. As it stands, it's always necessary to go through the motions of assigning Samples to a Partial, and then mapping Partials to a Patch.


The standard S770 comes with 2Mb of RAM, giving 20.7 seconds sampling time at a rate of 48kHz (in mono). With the maximum capacity of 16Mb, this rises to 167 seconds (almost three minutes). Extra memory can be purchased in the form of 2Mb SIMMs (Single In-line Memory Modules) which clip into the RAS770 memory expansion card inside the unit. Roland were originally asking £899 for the RAS770 board, but have since decided to fit it as standard. However, at £550, Roland's official OMS770 2Mb upgrade kit is ridiculously expensive, considering it consists of two standard 1 Mb SIMMs. These are identical to the SIMMs used in the Apple Macintosh computer, and as any Mac aficionado will tell you, 1 Mb SIMMs can be purchased for around £75 inc VAT - which makes a non-Roland memory upgrade some £350 cheaper!

Significantly, even when all the memory slots are filled, the RAS770 board still has spare sockets available, which suggests further RAM upgrades.


The S770 is a Fixed Sample Rate (FSR) device using Differential Interpolation. This weighty description basically means that it produces noise-free samples even when they are transposed to extremes. FSR has many other benefits too, including a real-time digital output. (See 'Differential Interpolation' side panel).

Figure 4.

As can be seen from Figure 4, the Sample page is fairly straightforward apart from the sampling frequency field, Freq, which offers 44.1/22.05kHz or 48/24kHz rates. This is because the S770's real-time digital output can operate at either 44.1 or 48kHz. Toggling between the two rates requires a trip to the System Parameters page to set the master sampling frequency, and it would be a good idea if this parameter could be duplicated on the Sample page. When a Volume contains Samples of mixed sample rates, the S770 converts them to the master sampling frequency at the digital output, though I have heard that Version 1 software has a slight problem with this.

Referring to Figure 4 again: Mode offers mono or stereo sampling. Time increments in 0.1s steps (automatically reducing the time 'Remaining' field), and Original Key decides, which key will be stored along with the Sample. The Input field determines whether sampling will be routed from the analogue or digital inputs. In the latter case, 'Digital Att' offers four fixed attenuation levels (+3, 0, -3, -6) which can be used to boost an under-recorded digital input signal. The front panel Sensitivity control and rear panel Gain Selection switch are used to set the optimum level, which is displayed in the lower half of the Sample page. If the last (red) segment is lit, the sample will be distorted.

When sampling analogue signals, Monitor On will route the input signal to the stereo and headphone outputs. The digital input is always present at the digital output (but not the stereo or headphone outputs), so Monitor has no effect when using the digital input.

Like Roland's S550/330, the S770 is continually sampling into the memory buffer allocated by the Sample Time field. This continual sampling offers two great facilities that other manufacturers would do well to copy. Firstly, audio threshold sampling need not lose the important front-end transients, because they are stored in a pre-trigger buffer (up to 0.1s). The pre-trigger is applicable to both Auto (audio threshold) and Manual sampling. '1 Way' sampling offers no pre-trigger facility but allows immediate sample playback, whereas all other modes take a few seconds before the sample can be played back.

The fourth sampling mode, Previous, is perhaps the most useful since it allows you to hear the sample before you take it: sampling ends when you click 'Start'! Again, this is possible because the S770 is continually sampling into the buffer set up by the Sample Time field, and in clicking Start you are simply deciding what to keep since it has already been sampled.

Whatever mode is chosen, clicking Ready brings up the Sample Execute page unless you've forgotten to name the Sample, in which case an alphanumeric keyboard appears on the screen giving you no choice but to name that Sample.

The Sample Execute page provides Start and Cancel buttons and retains the input level meter. Manual, 1 Way, and Previous mode Samples can also be forced from a MIDI Note-On or footpedal event (set up in the Trigger field).

A System Parameter can be set to automatically turn the internal fan off and on as the Sample Execute page is accessed. However, this still leaves the possibility of the fan being left off if you are distracted while in the Sample Execute page - an auto timer might be the solution.

After a slight delay (except with 1 Way sampling) the Sampling Over screen appears, offering an oscilloscope waveform display (one per channel) and various soft switches - Next and Retry take another Sample with the same settings, End and three more jump straight into the Loop, Truncate, and Normalise pages.


As with the Akai S1000, a newly recorded Sample can be played from a front panel button (Sound Play), so that Samples can be made without recourse to any MIDI device. However, if one is plugged in, the Sample is stretched across the full length of the keyboard, rooted at the original key.

Listening to a transposed Sample, the advantages of the S770's fixed rate Differential Interpolation sampling are immediately apparent - even the most extremely transposed sounds are clear, noise-free, and smooth. The Mickey Mouse effect is not overcome, but the gritty, grainy quality of variable sample rate machines is certainly eradicated.

In an A/B test, sampling digitally from a CD into both an S1000 and the S770, there was really no comparison. Both machines maintained the stereo image very well, but the S770 was less coloured and, in particular, its bass end was truly excellent. The S770 also scores over the S1000 in the dynamic range department, due to Roland's use of a 20-bit DAC.

Figure 5.


The Sample Truncate screen (Figure 5) displays the mono or stereo Sample allowing either side to be edited independently. The display can be zoomed in or out from x1 (which displays the entire Sample) to x64. Most commercial sample editing software provides multiple display markers which are essential for long samples. At present, the S770 offers three sets of markers in the form of From/To, Loop Start/End, and Release Loop Start/End points. (The latter two are accessed from the Looping page).

In the Truncate page, the section of the Sample to be played back is set up in the Key On field. In addition to playing back the Sample between the above points, this can also play back the portion of the Sample after the Release Loop point. (See Figure 6)

Figure 6.

Truncation usually leaves a Sample with a sudden cut-off, which is especially noticeable on ambient sounds. Emu's Emax and EIII have a facility called 'Taper' which allows the start or end of a sample to be digitally faded from a user-defined point. The S770 has a Fade function, which is a slightly cruder version, since it operates on both the start and end of a Sample with no control over the start point or slope. Nonetheless, it's a difficult function to live without, and future updates may well see an extension of this feature.

Another useful facility, first seen on the Emulator III, is auto backup, which automatically stores the truncated Sample under the same name, adding an 'N' suffix.

Two more features you would expect to find in the Sample Edit pages - copying and splicing - are in fact implemented as much more sophisticated functions in the S770's Resampling page, which we will cover a little later.


The S770 offers plenty of great looping facilities. One major improvement is that Samples do not need to be continually re-triggered in order to hear the effect of new loop parameters. This means that you can hold down a note or keep the sustain pedal pressed while changing the loop points. When 'Loop' or 'R.Loop' is selected as the Key On parameter, the loop data fills the graphic display. If Loop Fine is on, the loop crossover point is also displayed at maximum magnification.

The pitch of some samples varies naturally over time, and this often results in a loop which is slightly sharp or flat. The Loop page allows both sustain and release loops to be tuned over a semitone range to overcome this problem.

The Key On mode offers the same options as the Truncate page, but the sustain and release loops can be read in many different ways. Apart from the usual modes, Oneshot reads data from the start to the loop end point only. Fwd+One plays and loops normally, but loops an extra time when a key is released. Alt (alternate) reads the loop portion backwards and forwards, while RevOne (reverse one-shot) plays the sample from the loop end point to the start (ie. backwards). Finally, Rev (reverse) does the same thing but then loops the data from the loop start point to the start point.

These various modes provide a quick way to try out speech manipulation ideas and production effects, and a Length Lock feature makes it possible to experiment without losing the original loop points. However, it would be much more useful if the S770 (and indeed all samplers) could store loop markers independently from the Sample. In tapeless recording systems, edit points are stored this way (making the editing non-destructive). If this were implemented on a sampler, it would mean that one Sample could be looped in lots of different ways without the necessity to store multiple copies of the same data.

The S770 has no auto-loop facility as yet, but offers a form of crossfade looping called 'Smoothing', which crossfades from the loop start to loop end points. With an excessively long loop, this can result in an unwanted chorus-like effect; again, Version 2.0 promises to add more parameters to the Smoothing page.

The last Edit Sample facility is the now standard Normalise, which expands a Sample to its maximum level thereby ensuring maximum dynamic range. Like smoothing, normalising alters the raw Sample data and so both functions offer a backup option.


Having edited a Sample to perfection, the next stage should be to assign it to the keyboard. However, with the S770 you first have to set up a Partial, containing the Sample (and up to three others), in the Edit Partial page before it can be mapped from the Edit Patch pages.

Figure 7.

Edit Partial offers five pages: Common, SMT, TVF, TVA, and LFO. The Common page (Figure 7) is where up to four Samples can be chosen and independently tuned. A chorus effect can easily be set up by loading identical Samples into a Partial and slightly detuning them.

As with Roland's LA synths, the logic behind the Partial is that a completely new sound can be constructed by combining up to four sources. The owner's manual consistently encourages the user to assign attack, sustain, and decay portions of several Samples into a Partial to make a totally new sound. An example of the possibilities of this approach is provided by the Pitch Key Follow (KF) parameter. Each Sample in a Partial can be independently assigned one of 32 KF values (-8/8 to +8/8), which provides an alternative pitch key scaling centred around the original key. (See Figure 8.)

Figure 8.

In a Partial containing just two Samples, one a breathy attack sound and the other a sustained sound, the pitch of the breath Sample can be made to change in more subtle increments than semitones, which would certainly produce a more natural sounding result.

The remaining Common parameters affect the whole Partial and include course and fine tuning, volume panning, output assignment, and SMT velocity control (on/off). SMT stands for Sample Mix Table, and the SMT page controls how MIDI velocity will affect the level of a Partial's four Samples - in other words: velocity switch, velocity mix, and velocity crossfade.

Figure 9.

The TVA (Time Variant Amplifier) page provides a four level, four rate volume envelope which appears on the screen as a graphic and can be edited by dragging its square handles with the mouse. As can be seen from Figure 9, the TVA page also offers four velocity curves, time velocity sensitivity (ie. a higher velocity produces a faster attack), key follow (key number affects volume), and time key follow (note number affects attack rate).


With three filter modes (Low, High, and Band pass) and a Resonance (Q) control, the S770's TVF (Time Variant Filter) is essentially a subtractive synthesizer. A slow filter sweep picks out harmonics very well, but the resonance stops short of self-oscillation. As the filter sweep can be determined from a MIDI Controller (eg. Modulation wheel), the S770 can be used to perfectly mimic the kind of real-time analogue synth effects that are currently plastered over everything in the charts.

The Command menu of the TVA and TVF pages provides instant predefined templates - such as sweep up, sweep down (TVF), and organ, piano, brass volume envelopes for the TVA - which is a great idea.

The LFO page offers eight waveforms (with delay and sync) and several interesting features, including a detune parameter similar to the Emulator III's variation control which sets up a user-defined variation in LFO speed per voice - excellent for string sounds. The LFO output can control pitch, volume, and filter depth.

Figure 10.


In addition to mapping of Partials, the Edit Patch pages offer a host of other parameters, including MIDI channel, program number, polyphonic aftertouch, LFO depth, pitch bending, pan, and output levels. Mapping is made easy through the use of an on-screen keyboard graphic (Figure 10) and the fact that the note boundaries can be assigned directly from a MIDI keyboard. One slight oddity is that Roland's C4 note corresponds to Yamaha and Akai's C3. Strange?


The major reason for digitally combining two Samples into one (new) Sample is to save on polyphony, but the S770's resampling algorithms also provide massive scope for modifying and creating new sounds.

Figure 11.

The Resampling pages offer six special algorithms, each with its own depth parameter, which incorporate TVA, TVF, and ring modulation processing (Figure 11). Algorithm 1 processes two Samples through a TVF and TVA and then combines them, the depth parameter controlling the mix. Algorithm 2 combines two Samples then passes them through two TVFs and one TVA, providing a very steep filter cutoff. Algorithms 3 and 4 introduce ring modulation before passing the Samples through two TVFs and a TVA. Here, the depth parameter controls the amount of ring modulation - greater depth producing a more unpitched, metallic quality in the resultant Sample. Algorithms 5 and 6 allow one Sample to be filtered before ring modulation, which produces a more musically useful sound result.

Resampling can also be used as a way of thickening or chorusing a sound, by combing two identical sounds with a course or fine pitch offset. A delay of up to five minutes (!) can be applied before the second Sample is triggered, which can introduce even more movement into the Sample. It's a shame that a delay facility isn't included as a Partial parameter, since it would provide a polyphony-safe way of combining two Samples without the need to resample (which creates a new Sample).

Another reason for resampling is to save memory, since a lower resampling frequency can be selected. For many sounds, a 24/22.05kHz rate would suffice (particularly if the signal is being filtered anyway), and Roland offer an emphasis parameter to brighten up the top end response.

Roland impose no limit on the number of times a sample can be resampled and the S770's 24-bit processing ensures that the result always remains noise-free. The resampling process itself lasts as long as the longest sound being resampled and is initiated by pressing a key on a MIDI keyboard. The new Sample then appears across the keyboard.


Some of the S770 sounds supplied on the internal 40Mb hard drive are actually D70 samples, and some are derived from the S550. However, many are totally new. The drums and percussion voices are very impressive. All the Performance memories include lots of processed, velocity switched, and edited versions. The S770 can also read S550, S330, and W30 sounds directly, using a convert facility, with seemingly random survival of processing like TVF and TVA. S50 disks, on the other hand, are not usable with this convert facility and need to be loaded into an S550 first.

S550 CD ROM discs, such as OMI's USV1 and Club 50's Master Performance Series Volume 1, can also be converted and a new S770 CD will shortly be available from Invision.

For user storage, any SCSI-equipped hard drive, CD ROM, or erasable optical disk will suffice, though Roland are understandably keen to tout their own CD ROM and M07 Optical drive.

Sample dumping to and from an Akai S1000 proved successful (sustain and release loops remain intact) but painfully slow, particularly as stereo samples have to be transmitted one channel at a time. Sample dumping is definitely not recommended for samples longer than two seconds, unless you're going on a short holiday! Talking of travel, Roland have taken the very sensible precaution of providing a hard disk head parking facility to prevent head crashing.


The S770's nearest competitor will be Akai's forthcoming S1100, which is set to be launched in August. The S1100 matches the S770's 48kHz sample rate, 24-bit processing, 20-bit DAC, and realtime digital outputs. The S1100 also offers some neat production facilities that Roland would do well to copy. Firstly, in an attempt to break into that most conservative of markets - video postproduction - the S1100 incorporates SMPTE (Edit Decision List) triggering of samples. Secondly, a built-in DSP chip speeds up sample editing and also provides multi-effects such as reverb, delay, chorus, phase and pitch shifting - again, a great feature. Thirdly, it can be expanded to 32Mb. From a musician's point of view, the S1100's main advantage is sample library compatibility with the rest of the S1000 range.

The S1100 is expected to be about £1100 cheaper than the S770, so what does the S770 offer for the extra £1100? Well, firstly an internal 40Mb hard disk, which is absolutely essential for any stereo sampler. That takes care of at least £500 straight away. The remaining £600 can easily be accounted for by the S770's extra eight voices, then there's built-in SCSI and digital I/O (both options on the S1100) and resampling, of course.

The S770 is obviously a longterm project for Roland, and the potential for new facilities like DSP and direct to disk recording are certainly there, but unfortunately Roland are playing their update cards very close to their chest at the moment. Still, Roland have confirmed that Version 2.0 software (which should be available by the time you read this) will include time compression/expansion, direct waveform drawing, speedier operation, and new sound archival facilities. Roland would also do well to provide an audio triggering facility, as this is still one of the most popular methods of replacing samples on tape, because the timing of the original performance is preserved. An AES/EBU digital output would also do the S770 no harm.

Operationally, the S770 is lightening fast to use with a CRT monitor connected, though mapping and editing of Partials could be improved - a dummy Patch facility would make life easier. The availability of TVA/FVF templates illustrates that Roland appreciate the time-saving nature of shortcut features.


As I mentioned at the start, Roland are aiming the S770 at both the sampler and the production tool markets. Another serious competitor in the latter category is therefore the Apple Macintosh-based stereo Sound Accelerator system, from Digidesign, which now offers three software applications - Sound Tools (SMPTE cue list, DSP), Studio Vision (combined MIDI sequencer and stereo recorder) and Deck (4-track recorder with DSP).

The Macintosh has the advantages of a consistent user interface and the ability to upgrade hardware and software on a continually evolving basis. For example, Digidesign recently announced a new improved Sound Accelerator card for which existing customers can trade in their old cards. However, there is currently no software that turns the Mac into a serious multi-voice sampler and, of course, portability is 186th on the list of adjectives one might use to describe the new generation Macs.


For the pro audio studio, sound quality is the most important criteria, and as the S770 sound quality exceeds that of digital multitrack recorders and mastering machines, engineers need not be afraid of transferring an entire mix into the S770.

The S770 is currently making a strong impact in American studios, which bodes well for increased third-party support and future library sounds.

The Macintosh may soon be able to offer high quality sampling facilities and improved sound quality, and Akai's forthcoming S1100 will no doubt appeal due to its DSP and S1000 library compatibility, but Roland's S770 is available now, is already on its second software upgrade, and has no direct rival in terms of polyphony and sound quality.


S770 sampler £4800.
0MS770 2Mb memory upgrade £550.
M07 erasable optical £5225.
CD ROM player £1340.
RC100 remote control £250.

All prices inc VAT.

Roland (UK) Ltd, (Contact Details).


Frequency Response: 20Hz-20kHz (+0/-3dB at 44.1kHz)
Total Harmonic Distortion: 0.01% (A/D/A)
Residual Noise Level: stereo out (L/R) greater than 80dBm (all volumes: Max);
individual outputs (1-6) greater than 85dBm (all volumes: Max)
Polyphony: 24 voice
Sampling: stereo Differential Interpolation fixed sample rate
Data Format: 16-bit linear
Output D/A Conversion: 20-bit
Processing Accuracy: 24-bit
Sampling Rates/Times (mono): 48kHz: 20.7s (2Mb), 167s (16Mb)
44.1kHz: 22.5s (2Mb), 181.8s (16Mb)
24kHz 41.3 (2Mb), 334.1s (16Mb)
22.5kHz 45s (2Mb), 363.6s (16Mb)
Sample Processing: TVF Digital Filter (Low, High and Band Pass); TVA; LFO for modulation
RAM: 2Mb standard, expandable to 16Mb
Disk Drives: internal 40Mb drive; 2HD/2DD 3.5" floppy drive;
S550/330 file compatible (via conversion)
Inputs: rear L/R XLR, front L/R ¼" jack
Outputs: 6 polyphonic plus stereo out (¼" jack); Digital I/O;
coaxial/optical real-time input/output
Display: 64x240 dot backlit LCD; RGB CRT output 200x320 dot
MIDI: In, Out, Thru
Dimensions: 430(W) x 132(H) x 420(D)mm
Weight: 12kg (26lb 6oz)
RAM Capacity: 1 Volume, 64 Performances, 128 Patches, 255 Partials, 512 Samples
Hard Disk Capacity: 128 Volumes, 512 Performances, 1024 Patches, 4096 Partials, 8192 Samples



Until recently, the majority of sampling devices employed the Variable Sample Rate (VSR) method to transpose the pitch of samples. VSR changes a sample's pitch by altering the playback sample rate. For example, to transpose a sample recorded at 40kHz down one octave, the playback rate is reduced to 20kHz. Unfortunately, there are several undesirable side-effects associated with the VSR method. Firstly, as the sample rate is reduced, the bandwidth of the sound is also reduced - from 20kHz to 10kHz in the above example. Secondly, an anti-aliasing filter in a VSR device has to track the playback sample rate in order to block out high frequencies that fall into the audible spectrum as the sample's pitch is transposed.

In any device which incorporates a digital to analogue convertor (DAC), antialiasing filters are designed to have a so-called 'brick wall' response (ie. extremely sharp, fast cutoff) to avoid unpleasant ringing effects. This is difficult for a fixed frequency filter to achieve, but for a tracking filter whose cutoff frequency is continuously changing, it's nigh impossible! Inevitably a compromise is made, and sound quality suffers.


In a VSR system, a Sample recorded at 40.000 samples per second (40kHz) and transposed down an octave will have only 20.000 samples (20kHz) to describe it. In a Fixed Sample Rate (FSR) system, the sample rate is maintained. So in the above example, the transposed waveform will also contain 40,000 samples. The extra 20.000 samples need to be interpolated (estimated) in real time by the FSR sampling hardware/software.


In calculating the new waveform, some FSR systems use Linear Interpolation.

The S770 calculates its new samples by Differential Interpolation, a more intelligent system which analyses the original sample and attempts to calculate more accurate sample points. This results in a more natural sounding analogue waveform.

In addition to maintaining constant bandwidth and removing the need for tracking filters, the use of Fixed Sample Rate sampling (linear or differential) produces several very desirable spin-offs. Firstly, the sample transposition range is not limited by filters or hardware and so samples can be played back over a range of 14 octaves. Secondly, as the sample rate is constant, a real-time digital output can be provided which allows the sampler to be connected directly to a digital mixer.

Also featuring gear in this article

Browse category: Sampler > Roland

Featuring related gear

Previous Article in this issue

Beyer TGX Microphones

Next article in this issue

The Art of Propaganda

Publisher: Sound On Sound - SOS Publications Ltd.
The contents of this magazine are re-published here with the kind permission of SOS Publications Ltd.

The current copyright owner/s of this content may differ from the originally published copyright notice.
More details on copyright ownership...


Sound On Sound - Jul 1990

Gear in this article:

Sampler > Roland > S770

Gear Tags:

16-Bit Sampler

Previous article in this issue:

> Beyer TGX Microphones

Next article in this issue:

> The Art of Propaganda

Help Support The Things You Love

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!

Donations for July 2024
Issues donated this month: 14

New issues that have been donated or scanned for us this month.

Funds donated this month: £20.00

All donations and support are gratefully appreciated - thank you.

Magazines Needed - Can You Help?

Do you have any of these magazine issues?

> See all issues we need

If so, and you can donate, lend or scan them to help complete our archive, please get in touch via the Contribute page - thanks!

If you're enjoying the site, please consider supporting me to help build this archive...

...with a one time Donation, or a recurring Donation of just £2 a month. It really helps - thank you!

Small Print

Terms of usePrivacy