Yamaha's new flagship workstation mixes 'n' matches several of the company's existing synthesis technologies and throws in a new one for good measure. But have they over‑extended themselves? Simon Trask spends some time with the EX5 to find out...
These days, it seems that if you're a company on the cutting edge of synthesizer development, it's no longer enough to provide just one method of synthesis. Korg, it could be said, have seen to that — first with the Prophecy and more recently with the Z1, while the Trinity workstation's sample‑based synthesis environment can be augmented by the Prophecy's multi‑synthesis capabilities, courtesy of an add‑in board. Yamaha aren't altogether new to this mix 'n' match game, having combined AWM and AFM synthesis as far back as 1990 with the SY77, before finally giving FM the heave‑ho with the SY85. More recently, Yamaha have taken AWM and VL synthesis — previously only available in two separate modules, the MU90R and the VL70M — and put them together in the MU100R. With their new EX workstation synthesizer range the company are taking things a stage or three further by integrating AWM, VL, AN (as on the AN1x) and new FDSP (Formulated Digital Signal Processing) synthesis in one instrument — essentially combining familiar sample‑based subtractive synthesis technology with several forms of virtual modelling.
The new generation of multi‑synthesis instruments imply an acknowledgement, at least tacitly, that no one method of synthesis is up to providing the full range of sounds that musicians want — even expect — in the late '90s. Of course, these days, the technology is powerful enough and at the right price to make the multi‑synthesis approach feasible — or is it? When Yamaha combined AWM and VL synthesis in the MU100R, the VL component lost its own independent multi‑effects processing, a significant component of its sound, while the new EX series makes some sonic compromises of its own. So does Yamaha's new synthesizer range fulfil the promise of a truly versatile and rounded multi‑synthesis sonic environment, or does it just dress up familiar sample‑based subtractive synthesis with a few sonic baubles and bangles?
There are three instruments in Yamaha's new EX range: the 61‑key EX7, the 76‑key EX5, and the EX5R rackmount MIDI module. For this review we're looking at the flagship EX5 model, but in large part my observations can be taken to refer to the other two models as well. Aside from the keyboard and a fair amount of weight and size, the EX5R merely loses a few foot‑controller inputs, MIDI B In and Out sockets, and the EX7's array of Bank and Program select buttons. The EX7 makes some more significant compromises for its cheaper price: half the polyphony, no VL synthesis, fewer synthesis combinations, less flexible Insertion effect routing, a single A‑D input, and no MIDI B In and Out or pair of individual outs (though it has the same expansion capability as the EX5/5R).
The EX5 is a solid, weighty instrument with a spacious, well‑organised front‑panel layout, six chunky infinite rotary knobs that just beg to be twiddled, and a modest backlit graphical LCD which I found rather disappointing after the Technics KN5000's glorious full‑colour LCD screen (see review in last month's issue). The 76‑key keyboard has a moderate travel and a pleasing semi‑weighted feel, though personally I could have done with a little less key bounce on the rebound.
Yamaha's new flagship has the familiar workstation components of multitimbral sound section, built‑in multi‑effects processing, onboard multitrack sequencer, and built‑in floppy disk drive. In line with the current vogue, it also adds an arpeggiator.
At the heart of the instrument's synthesis capabilities is an AWM sample‑based synthesis section. However, the EX5 also augments its 16Mb of sample ROM with built‑in stereo sampling capabilities, though what you won't find is any facility for multitrack digital audio recording, a la Trinity — the EX5's sample RAM is geared towards expanding the source samples for AWM synthesis. At the same time, the EX5 also goes beyond familiar sample‑based subtractive synthesis by incorporating the Virtual Acoustic synthesis of Yamaha's VL‑series synths and the Virtual Analogue synthesis of the AN1x, along with a new synthesis technology called FDSP, which actually provides multiple synthesis and effects processing methods that can be applied to the output of the AWM synthesis section.
One thing the EX5 isn't is an XG or GM playback instrument, despite its 'Sondius XG' label (which, confusingly, refers to the fact that it incorporates VL synthesis, which is classified as an XG extension). XG and GM song files won't play back in any meaningful way, then; however, Yamaha will be bringing out a promotional CD containing a GM sound set for the EX series.
The EX5 is so versatile sonically that it would be impossible to cover fully here.
As touched upon earlier, the EX5 utilises not one but several synthesis methods. At its core is Yamaha's familiar AWM2 sample‑based subtractive synthesis architecture, which for its sample sources draws on a 16Mb sample ROM containing almost 2000 samples and waveforms. This impressive array of sounds can be augmented by sounds sampled via the EX5's A‑D inputs or uploaded via SCSI into the synth's sample RAM. In fact, you don't even have to bother with sampling or uploading external sounds, as the EX5 lets you sample its own effected output internally; you can sample anything from single notes to keyboard performances and multitrack sequences, and of course then use that sample as a sound source to be synthesized, effected and sampled again, through multiple generations if you want. Not that the measly 1Mb of sample RAM provided by Yamaha is going to get you very far, so if you're going to take advantage of this aspect of the instrument at all, I'd say it's worth investing in extra sample RAM from the outset (which can be both SIMMs and 8Mb of Flash memory). You can store up to 1024 samples in the RAM, so this is no half‑hearted additional capability. Bear in mind that you'll also need a SCSI‑compatible external storage device to hold all your samples.
Sampling on the EX5 is a straightforward process. You can sample in mono or stereo at 16‑bit, 44.1kHz resolution. The EX5's sample editing capabilities are by no means extensive, however; essentially, you get basic graphical looping facilities and copy, delete, append, normalise and extract functions — don't even think about time‑stretching or pitch‑shifting. There's no scrolling cursor bar for the graphical sample display, to help you pick out sample points, and you have to re‑trigger the sample each time you select a new loop point. You can scroll numeric start, loop and end fields in 10s, 100s or 1000s for speed on the looping page, but only from the controller knobs, not the value dial — and not at all for start and end sample values on the Extract function page. Basically, if you're serious about using custom samples you'll need to transfer them to external sample editing software (the EX5 supports SMDI transfer of samples via SCSI). Of course, you can also load in existing samples from other sources (the EX5 supports WAV, AIFF and Akai formats).
The EX5's familiar AWM synthesis architecture of oscillator, filter and amplifier, with associated EGs and two assignable LFOs, is fully programmable, and offers a satisfying depth of functionality. When you assign a sample from the EX5's sample RAM to the oscillator, a Wave Edit option appears in the LCD, above the F3 button; press this and you go to a page where you can define a keymap with up to 128 samples. You can mix and match samples from ROM, RAM and Flash memory, and give each sample its own key and velocity range, along with other settings such as level, coarse and fine pitch, and pan (no random pan option, though); note that no more than two samples can sound simultaneously on a key.
The filter section has one resonant Static Control Filter, followed by two resonant Dynamic Control Filters which can be configured in parallel or in series. The SCF offers a choice of eight filter types (low‑pass, high‑pass, band‑pass, inverted low‑pass, parametric EQ, and several shelving filters), while DCF1 offers eight types (low‑pass 24, 18, 12 and 6dB, high‑pass, band‑pass and band elimination) and DCF2 offers four (low‑pass, high‑pass, band‑pass and band‑elimination); DCF2 is only active if DCF1 has a setting other than low‑pass 24, low‑pass 18, or Thru. The filter EG applies to both DCFs, while velocity control of both filter cutoff and resonance is available. Pitch and filter EGs provide an initial hold time, plus six stages, while the amplitude EG provides seven stages; pitch and filter EGs both offer a choice of three options for envelope looping prior to release. The amplifier stage includes a pan position parameter (L63‑R63) with random, key follow and random depth options.
LFO1 offers a choice of sawtooth, triangle or square waveform, plus sync on/off, delay time, fade‑in time, speed, velocity‑sensitivity amount, amount of random variation in LFO speed, and separate amounts for pitch, filter and amplitude modulation depth. Meanwhile, LFO2 adds sawtooth down, trapezoid and two Sample & Hold waveforms, along with LFO phase and modulation of LFO1 speed and pitch, filter and amplifier mod depth.
These days, it seems that if you're a company on the cutting edge of synthesizer development, it's no longer enough to provide just one method of synthesis.
All these parameters go to make up a single AWM Element. An EX5 Voice can have up to four such Elements, which can be both key‑ and velocity‑split/layered on the keyboard, and can, of course, have their own pan settings. Each Voice also has a sophisticated modulation matrix which allows any 16 of up to 34 AWM parameters to be modulated dynamically from any of 13 mod sources (pitch‑bend, aftertouch, the two EX5 mod wheels, foot controller, ribbon controller, breath controller and the six front‑panel controller knobs); velocity control of various parameters is programmed within the synthesis architecture. Basically, you can define up to 16 Controller Sets for each Voice, with each Set consisting of one destination and an associated mod depth setting, plus on/off settings for each of the 13 sources; you can also turn each Set on or off for each of the four Elements.
Finally, the EX5's built‑in effects processing provides a reverb processor with a choice of 12 effects, a chorus processor with a choice of 17 effects, and two Insertion effect processors with a choice of 24 and 79 effects respectively — all programmable for each Voice. Each Element can be routed to Insertion effect 1 or 2, or can be set to bypass the Insertion effects, while separate chorus and reverb send levels can be set for the output from the Insertion effect(s), together with any direct signals. Chorus and reverb send amounts, plus parameters from Insertion effects 1 and 2, can be selected as modulation destinations within the Controller Set matrix.
As you've probably realised by now, the EX5 is no lightweight when it comes to programmability — even with 'just' AWM and the synth's open‑ended sample source capability. However, as mentioned earlier, the synth can also call on three other synthesis methods: AN, VL and FDSP.
Each EX5 Voice can be set to one of seven Voice Types, or configurations of synthesis methods. With the exception of the Drum mode (see page 176), each of these options has four Elements (though you can turn off individual Elements if you don't want to use four), which can be key and velocity split/layered and routed through the reverb, chorus and insertion effects. The Voice types are as follows:
VL + AWM: a VL sound can be assigned to Element 1, while Elements 2‑4 can be assigned AWM sounds.
AN (Poly) + AWM: an AN sound can be assigned to Element 1, while Elements 2‑4 can be assigned AWM sounds.
AN (Layer) + AWM: assigns an AN sound to each of Elements 1 and 2.
AN + FDSP: assigns an AN sound to Element 1 and an FDSP sound to each of Elements 2‑4.
The VL Element is monophonic, as on the VL70M and MU100R, while the AN Element is duophonic in the AN (Poly) + AWM mode and monophonic per Element in the AWM (Layer) + AWM and AN + FDSP modes.
AN synthesis is, of course, Yamaha's ANalogue modelling synthesis, as introduced on the AN1x, only here without the AN1x's polyphony — which means that its application on the EX5 is orientated towards bass and lead sounds. The AN1x's synthesis architecture is carried over into the EX5 (though without its noise source, seemingly), so you can program your own AN sounds. A total of 42 AN parameters are available for dynamic control in the Controller Set modulation matrix, while velocity is added as a mod source for the AN Element(s).
For the VL Element in VL + AWM you can select one of 271 preset VL 'waves' and route it through a synthesis architecture of 5‑band parametric EQ and low‑pass resonant Dynamic Control Filter, with pitch and amplitude envelopes and LFO modulation. You can't edit the VL driver, pipe/string and modulator parameters directly, but you can select some of them, such as embouchure, tonguing, throat and growl, as mod destinations in the Control Set matrix. FDSP is covered in the 'FDSP Synthesis' box below, so I'll pass over it here.
Drum mode is for creating drum kits containing up to 128 AWM sounds, each of which is assigned to a specific key and has its own level, pitch, pan, reverb send, chorus send and Insertion effect 1/2/off settings. You can also layer multiple samples on each key, up to 64 drum layers (!), and assign a velocity range and crossfade amount for each layer; each of these layers also has its own drum parameter settings, as above.
The EX5 is a solid, impressively accomplished professional synth.
In addition to its Voice memories, the EX5 also has Performance memories, which allow up to 16 Voices to be used in keyboard‑split/‑layer and 16‑part MIDI multitimbral configurations. Reverb and Chorus effects are programmed globally for each Performance, with individual Parts having their own programmable reverb and chorus send levels, while the Insertion effects are those programmed for the Voices assigned to the Parts. Don't get too excited, though: you can turn Insertion effects on for up to four Parts if all 16 Parts use AWM‑only Voices, while if an AN, VL or FDSP Voice is used that number comes down to one Part. What's more, you can't assign more than one AN, VL or FDSP Voice to a Performance (so, for example, if you have an FDSP Voice assigned to a Part, you can't assign another FDSP or an AN or VL Voice to another Part). It seems the DSP can't handle more than one of these synthesis types at a time, or more than one Voice at a time with the same non‑AWM synthesis type; this means that in a multitimbral setting the bulk of the work is still going to be done by AWM‑only Voices. If you try to select a second FDSP, AN or VL Voice a window pops up in the LCD with the message 'DSP resource full! Alternate voice selected', and the EX5 automatically selects the next AWM‑only Voice. This quickly becomes annoying when you're scrolling through the Voices trying to find a suitable one for a Part.
As well as its Pattern‑ and Song‑based sequencing capabilities, the EX5 also features an arpeggiator. Today's digital synths are transforming this once‑humble device into a much more sophisticated and complex beast — which may or may not be missing the point. Yamaha are evidently trying to outclass everyone else with the EX5's arpeggiator, turning it into a 4‑track 'live sequencer' with all the recording and editing functions you'd expect from a workstation sequencer.
There are 50 preset and 50 user arpeggio memories. Each Voice can be assigned any one of these arpeggio memories, and you can also program default arpeggio on/off status, arpeggio tempo and arpeggio trigger key range per Voice, and assign one of the six controller knobs to control the tempo live. Pre‑recording a sequence of notes may seem an odd thing to do with an arpeggiator, where the notes are defined at the 'playback' stage. However, the EX5 treats recorded arpeggiator notes not as pitches per se but as octave numbers and key numbers relative to C. These number sequences, in conjunction with the selected note search mode (for example, 'note up 1 oct') and the actual notes that you play on the keyboard, determine the note sequence played — whether the arpeggio trigger notes are played as a chord or individually, whether they are played up or down in sequence, and whether some notes are transposed into different octaves. Recorded note velocities and durations can also be superimposed on your trigger notes. Now imagine that you can have four tracks going on simultaneously, each 'superimposing' its own note sequencing on the arpeggio trigger notes! You can get some pretty complex live sequences going, all triggered from a few held notes. Probably the best way to create something usable is to record each track in a different octave, so that the results play back in different octaves.
Arpeggio tracks can also transmit via MIDI, with each track having its own MIDI channel. Track recording can be in real or step time, with replace and overdub options for real‑time recording, and you can include pitch‑bend and mod‑wheel data. Arpeggio tracks can be edited at event and bar levels (copy, clear, append, quantise, and so on). You can also superimpose groove templates on each of your tracks within an arpeggio, selecting from 100 presets or creating one of your own.
In Pattern sequencing mode you can record up to 50 8‑track Patterns in real‑time or step‑time modes, with the same event‑level list‑editing capability and bar‑level editing functions as the arpeggiator and the Song sequencer. Patterns, which loop in record and play modes, can be up to 16 bars long, but a neat feature is that each track can have its own length. The longest track defines the overall Pattern length, but each track loops according to its own length. You can define lengths to quarter‑note resolution, not just whole bars, with the minimum length being one quarter note. Each track can be assigned to transmit on any MIDI channel via Out A or B. You can also set Play Effects for each track of each Pattern; these provide features such as groove template, clock offset, and gate and velocity offsets.
The EX5's multitrack Song sequencer provides 16 linear tracks, a Play Effects track, a Pattern track and a tempo track. In the Pattern track you record Pattern numbers in real time; the EX5 switches to the selected Pattern at the beginning of the next bar. You can also edit Pattern track data within Song mode. The Play Effects track is used to record the various playback modifications associated with each Pattern. You can also record Play Effects for each of tracks 1‑16.
For performance recording, Song mode has step, multi, replace, overdub and punch‑in/‑out record modes. You can start and stop recording at any bar location in the sequencer, and delve into event‑level editing or draw on 24 bar‑level editing functions. All in all, it's a very capable workstation sequencer.
The EX5 is so versatile sonically that it would be impossible to cover fully here, so I'll limit myself to general impressions and one or two specific examples. One general point worth making about the EX5's overall sound is that it is richer and fuller and has a more professional sheen to it than the sound of Yamaha's MU‑series XG instruments (and of course, unlike them, it's fully editable). The EX5 assigns its Voices to 22 instrumental categories, which says something about the breadth of its sonic coverage. Acoustic and electric pianos are up front in the Presets list, and Yamaha are as strong as ever here, particularly on the electric pianos (the acoustics do 'pop' better than 'classical', to my mind). Pads and strings are a particular strength; Preset 1:G03: 'EnsembleMix' is a fine example of a wonderfully rich, smooth analogue‑style string sound (though it's actually a 3‑Element AWM sound — remember, AN synthesis on the EX5 can't do lush chordal parts). The many strong basses and leads benefit from the inclusion of Yamaha's AN analogue synthesizer modelling technology; the EX5 really punches and booms at the bass end, and manages to be both rich and penetrating in the lead department. Meanwhile, P1:E16: Jazz Trump is a fine example of a realistic, naturally responsive trumpet sound that just couldn't be achieved with sample‑based or analogue‑modelled synthesis. A good example of how FDSP modelling can enhance AWM‑sampled sounds is Voice Preset 1: A14: Michel, which routes four AWM electric piano samples through the FDSP EP Pickup model. The result is a more naturalistic and 'breathing' electric piano sound, both richer and mellower, than the same Voice without the FDSP processing. See the 'FDSP Synthesis' box for another example of how FDSP can enhance the familiar AWM sound world.
The EX5 is a solid, impressively accomplished professional synth which manages to combine familiarity with originality and versatility in an authoratitive manner. A characteristically Yamaha instrument in both presentation and sonic character, even with the new FDSP technology, it will appeal to anyone who likes the company's typical mix of sonic brightness and warmth, power and delicacy, deep bass and piercing highs. At the same time, the bringing together of AWM, AN and VL synthesis, the introduction of FDSP and the inclusion of sample RAM make the EX5 an instrument of true sonic versatility, and I can see it finding a place in all manner of musical applications — though the fact that you can't use more than one FDSP, AN or VL Voice in a multitimbral setting is frustrating. More an evolutionary than a revolutionary instrument, the EX5 is in some ways less than the sum of its parts, in other ways more.
|VL + AWM||1 + AWM||–|
|AN (Poly) + AWM||2 + AWM||1 + AWM|
|AN (Layer) + AWM||1 + AWM||–|
|AN + FDSP||AN: 1; FDSP: 8||–|
FDSP, or Formulated Digital Signal Processing, to give it its full name, treats the note output of one or more AWM Elements as an input signal for further synthesis (Yamaha call FDSP an "adjunct" to AWM synthesis). When you select FDSP as the Voice type, the Voice has all the sample‑based subtractive synthesis capabilities of standard AWM mode, but in addition allows you to route the four available AWM Elements through any one of 10 FDSP synthesis types:
- 01: Electric Piano Pickup
- 02: Electric Guitar Pickup
- 03: Water
- 04: Pulse Width Modulation
- 05: Flange
- 06: Phaser
- 07: Self FM
- 08: Tornado
- 09: Ring Modulator
- 10: Seismic
While the straight AWM mode gives you 126‑voice polyphony, an FDSP Voice has only 16 voices of polyphony to draw on; the polyphony takes a hit because a lot more processing power is required to perform the FDSP processing in addition to AWM synthesis. The EX5 and EX5R, but not the EX7, also have an AN + FDSP Voice mode which replaces one of the AWM Elements with a monophonic AN Element, though the AN Element isn't available for FDSP processing; also, polyphony in this mode is further reduced to eight voices.
You can turn FDSP routing on or off independently for each AWM Element, allowing you to combine straight AWM and FDSP‑processed AWM sounds within a single Voice. A good example of this is Voice G16: Abendstern, from Preset Bank 1, which overlays smooth choral AWM sounds with tinkling bell‑like ripples, to create a rich, shimmering pad sound reminiscent of Korg's Wavestation. The bell‑like sequence is created by routing a sawtooth waveform‑based AWM Element through an FDSP type called Water.
It's worth emphasising that FDSP processing is part of the individual voice synthesis architecture, so FDSP effects such as flange and phaser are note‑specific; velocity and
key‑follow modulation of selected parameters within each type allow for shaping of sounds per note.
Element on/off setting, FDSP type selection and FDSP parameter editing are all done from a single page, and parameters range from 15 to 21 in number, depending on the FDSP type. The Cursor up/down buttons and value dial or the controller knobs are used to select and edit FDSP parameters within a scrollable window; the controller knob method, in particular, allows for quick and easy editing of the parameters.
|Keyboard:||EX5: 76 keys, EX5: 61 keys; velocity and channel aftertouch|
|Sound Generation:||AWM2, AN, VL (not EX7) FDSP, sampling|
|Polyphony:||EX5/5R: 126 voices; EX7: 64 voices|
|Memories:||512 (256 preset, 256 user) Voice; 128 user Performances|
|Sample RAM:||1Mb (standard), up to 64MB (2 SIMM slots), 8Mb Flash RAM optional|
|Sampling:||16‑bit linear, 44.1kHz stereo|
|Effects:||Reverb (12), chorus (17), Insertion (79); 2 Insertion effect processors|
|Song Sequencer:||16 tracks, 480ppqn resolution, approx 30,000‑note memory|
|Pattern Sequencer:||8 tracks, 50 patterns|
|Arpeggiator:||4 tracks, 50 preset memories, 50 user memories|
|Display:||64‑ x 240‑pixel backlit LCD|
|Floppy disk drive:||3.5‑inch DSDD/HD|
- 76‑key keyboard.
- Heavyweight look and feel.
- Plenty of front‑panel buttons, plus six infinite rotary edit knobs.
- Rich, professional sound.
- Provides several synthesis methods, plus sampling.
- Large sample RAM capacity.
- SCSI port for mass storage.
- LCD interface can seem cramped.
- AN synthesis has less polyphony than the AN1x.
- Insertion effects on only one Performance part when an AN, VL or FDSP Voice is used.
- Only one FDSP, AN or VL Voice can be used in a Performance, Pattern or Song.
- Selecting Voice or Performance Edit mode cuts short any active notes.
Yamaha have produced a solid, powerful and impressively versatile synthesizer which is well suited to both stage and studio settings.