Samplitude from Magix is now one of the more established names in the music technology market place, with a growing and dedicated band of followers, especially in Europe, using it for recording, editing, mixing and mastering. This growth might, in part, be a trickle-down effect from the recent rise in the popularity of Sequoia, Samplitude's bigger — and very much more expensive — dedicated mastering cousin. Even just a couple of years ago, Sequoia was still being described in professional circles as a 'relative newcomer', as an upstart challenger to the then worldwide dominance of Sonic Solutions and SADiE systems, but it now boasts many top-rung mastering engineers as recent converts. My own facility had been a loyal SADiE supporter for many years, but then the narrowing gap of functionality (especially in terms of classical editing capabilities) and an increasing gap in price made us look again at native-based editing technology, and when we came to think about replacing our SADiE 4 system a year or so ago we too took the plunge for Sequoia, keeping SADiE in commission mainly for its suite of CEDAR noise-reduction processors.
Even without its family connections with Sequoia, Samplitude has been gathering a reputation for offering serious capabilities coupled with tremendous value for money. It upped the free plug-in ante, for example, when it became one of the first software systems to offer free POW-R dithering (very much the dithering option of choice) giving expensive stand-alone dithering systems a jolt. Its range of internal processors is very highly thought of, especially its Room Simulator and phase-coherent Multi-band Dynamics, and it has also been getting good press for workflow-related features such as rapid CD burning and secure back-ups, which are important considerations in a professional environment.
But all software evolves, and much of what was unique 18 months ago is ubiquitous today. So can Samplitude in its new incarnation — version 9 — still compete?
Samplitude 9 certainly seems to be keeping up the pressure on its competitors with its bundled plug-ins. These may not be as extensive, or as visually stunning, as those found in some other programs, but they are carefully chosen and, in the main, very nicely executed. From this year's crop, for example, I found good use during the review period for a very musical basic compressor and an 'optical' compressor from the Analogue Modelling Suite, a useful reverb from the Vintage Effects Suite, a De-esser and a Spectral Cleaning plug-in which offers — for free — something similar (in broad principle anyway, if not exactly in practice) to the very expensive Algorithmix Renovator and CEDAR Retouch.
The new version of Samplitude also contains literally dozens of other new features: improvements to many of its existing plug-in effects; the addition of a whole suite of new signal processors; additional MIDI capabilities, including a new Score Editor; more effective and ergonomic track handling and file management; and a new Hybrid Audio Engine which, when working with ASIO drivers, allows a sharing of system resources to optimise latency/performance trade-offs.
It would take up a large part of this review simply to list and describe all the new features in any detail. So what I will do instead is to describe how we concentrated on just some of these new features, and a couple of the old ones, to show what was possible in a week or so of music production from a moderately priced, all-in-one-box, software solution. In one sense this was a 'real-life' test as all of the sessions documented here were professional productions whose results were judged, not only by us, but also by the artists and the record companies paying for them. In another sense it was not at all realistic, as no one system is ever considered sufficient in itself to perform all of the tasks to an optimum level: specialist plug-ins and outboard processors are used as a matter of necessity and a matter of course. So, the fact that at some points we had to move outside of Samplitude to complete a production is not then necessarily a criticism of it, but the fact that we could work within it for as long as we did is certainly praise.
In The Spectral Wash
Unlike the included POW-R plug-in, which is the real thing, Samplitude's Spectral Cleaning, despite visual similarities, is not Renovator. We found that it can do a pretty decent job of removing very discrete and specific transients, and can usefully diminish some slightly more smeared noise, but beyond that, in the exposed situation of most 'classical noise', it was not up to full professional standard (which is, of course, no real criticism of a free-plug in when the real thing costs in the region of £2,000). So our use of it on the Karine Georgian session was not as an absolute removal tool, but simply as a guide to necessary retakes. If, during recording, we heard a noise, we immediately tried some quick-fix noise removal on it, and worked on the basis that if we couldn't improve it at all with Spectral Cleaning then we'd better do another take. If, however, we could at least approximate a decent cure, then we knew we would more than likely be able to do much better later with Retouch. That indeed turned out to be the case: the left-hand screen (below) shows a portion of audio material that had a sharp non-musical bow noise (the central spike) superimposed on a genuine note (the broad horizontal portions continuous on either side). Having isolated that spike (right-hand screen) and Spectrally Cleaned its upper portion we were convinced that Retouch would easily be able to remove it. And it did.
Samplitude is still a Windows-only application, and Magix recommend a minimum CPU speed of 1.5GHz, and 1GB of RAM, for running Samplitude 9 on Windows XP. That means that the host computer on which we installed the software slightly exceeded that specification, but it is still a pretty modest setup by current standards of what is available. It's based around a 3GHz Pentium 4 with an 800MHz front side buss, and 1GB of RAM, and was put together as a recording rig for us by DACS over two years ago; in their custom black anodised rackmount livery, it looks like Darth Vader's desktop.
The soundcard we used for these sessions was a Lynx AES16, and we simply let the Samplitude software run with its default settings, using MME drivers. (ASIO is of course the preferred option for low-latency operation when using third-party plug-ins or running external effects looms, but as we intended to keep all processing within the Samplitude system, we decided to use the vanilla Windows drivers.)
Installation, from a DVD which also includes tutorial demos and the like, was a breeze, with all the drivers doing their job and talking to each other with no fuss or problems whatsoever.
After running briefly through some of the new features to begin to familiarise myself with them (and to have fun: a couple of the settings in the new Ecox and Filtox plug-ins made wonderful chaos of some rather sober string quartets), I then used the program to complete three very different projects.
The first was a high-resolution recording of some classical music — the Russian cellist Karine Georgian playing Bach's Suites For Solo Cello — which took place on location in St Martin's church in East Woodhay (a favourite venue for recordings that require a warm, but controllable acoustic). The second involved editing (vocal comping), minimal processing and mixing of some tracks for a CD by a young female vocalist; and the third was mixing the first CD from the acoustic jazz trio Quiet Focus.
The common element in all of these productions was simplicity of instrumentation and hence open texture, making them an ideal testing ground because of their intolerance of any kind of distortion. Heavy or dense material might have masked the subtle degradations that digital devices can introduce into the signal path; but in this music, everything was very much open to critical view.
Karine's performance of the entire set of Bach's six Suites is due for release in 2007 and this was the second of the three sessions that have been planned to record the material. The project is an 'audiophile' production and so the signal path was minimal and absolutely pure: a pair of Sonodore RCM 402 omni mics, a Grace 801 preamp and a Mytek A-D converter operating at 24-bit/96kHz, all captured on this occasion by Samplitude 9 via the Lynx digital interface card.
In such situations, capturing hopefully inspired performances, and where the technology has to be entirely 'invisible' to the performer, the demands on the recording software are many and varied. Of course, its main task is that it has to be able to capture with no loss or degradation the output of the A-D converter; but in classical location work it also has to be a little more. The system has to be quickly responsive and allow for easy organisation too: the producer, or the musician, can call for a new take with very little warning, so the engineer has to make sure that his recording system can respond immediately — and it helps enormously if it can also provide foolproof, automatic ways of logging and organising the many takes and part-takes that are a regular feature of modern classical music production.
I like to work in hour-long file 'reels' (which are then easily run off as audio on CD, and to DVD for back-up) and it is not unusual for such a file to contain 50 or 60 takes, so Samplitude's automatic take numbering was a real boon. At the end of each take I pressed a single button to close the recording and save it, and then the press of another button brought the software back to a state of readiness for the next take, complete with 'reel' and incremented take number. In Samplitude each take creates a new Object, so when we took a break at the end of the 'reel', I then used the 'Set CD Index to Object Edge' and very swift CD burning functions to make an audio CD of that part of the session. Even though burning 'on the fly' was not possible because it involved sample-rate converting from 96kHz down to 44.1, the whole process was so fast and entirely trouble-free that I could produce a CD of what we'd just rcorded even before everyone else had finished their cups of tea!
Another feature of Samplitude that I found extremely useful during these sessions was the feedback that is provided by its various 'visualisations'. Of course we have to rely on our ears in the final analysis, but it's nonetheless a considerable advantage to be able to check what they are telling us against the objective picture given by visual metering. Samplitude's phase and correlation meters have obvious uses, as do the peak level meters, but we also found an unusual use for the Spectrogram — the visual representation of the material being recorded on a moving timeline which 'draws' its frequency and amplitude creating a screen full of colour which can be read almost as an orchestral score. (See the screenshot overleaf, which shows the fundamental of the solo cello line, traced in parallel above it by the harmonics). Because there is barely anywhere in England that is not within earshot of a flight path, our sessions were occasionally interrupted by overhead planes. As soon as we heard them we stopped, and as soon as we couldn't hear them we started again. But it began to be a strain and a bit of a pain when we couldn't quite agree whether the noise had faded far enough to begin again or not — and here the Samplitude Spectrogram came into its own. The Sonodore mics have an extremely broad frequency response and are very sensitive, so even when there was 'silence' in the church, the Spectrogram showed us visually the physical noise floor (for example, low-frequency rumbles below 40Hz down at 50dB, which are clearly visible in the screenshot below the main musical action). So instead of wasting time discussing whether the plane noise had abated sufficiently or not, we simply looked at the screen: if it showed that there was anything more than the floor, we waited.
Generally it is the demand for swift response and rigorous solidity that sorts the wheat from the chaff in location recording equipment, and until now has been one of the main motivations for choosing non-native systems (such as Pyramix or SADiE) which, being run on dedicated DSP cards, are widely regarded as inherently quicker and more stable than programs that depend on native computer power. If that were ever completely true, it seems now to be changing: we ran Samplitude for 10 hours a day for four days (and have subsequently run it at the same intensity) with many user interruptions, and regular multitasking, and it didn't suffer from a single glitch. It's also worth noting that native systems tend to run more quietly: this is not a huge consideration in a studio environment with a separate machine room, but occasionally crucial on location where there is usually no luxury of a separate room (and sometimes not even the luxury of a separate control room). I regularly meet Pyramix engineers who have to run lines to their monitoring space from their main systems, which have to be left in their vans outside.
On the other hand, I still have to sound one sour note: the enforced use of a USB dongle to confirm the continued legitimacy of the installed software is, for the location engineer, a disaster just waiting to happen. Such a setup is fine when the computer is safely installed in a studio rack, but when it is regularly moved, there's a real possibility that the flimsy key could be damaged in use or in transit, as well as the likelihood of do-do moments when the engineer actually forgets to take the dongle with him. I tend to barricade my computer away from possible dangers, but it's far from ideal, and I really wish I didn't have to.
Our second project for Samplitude was post-production work on a few tracks from a non-classical CD by a young female vocalist. Most of the tracks featured harmony vocals and a full band backing, but on some songs her solo voice was accompanied by just piano, or a stark arrangement of strings. Although there was no requirement that the music be absolutely 100 percent naturalistic — imagine an English Tori Amos — the producer called us in as he felt that the vocals on these tracks needed to be treated with much more delicacy. 'Treatment' here was largely a euphemism for 'bringing back into tune': even the loveliest of voices can stray from the centre of the note. Although moderate use of modern auto-tuning software is generally fine against the backdrop of a dense mix, vocal comping — the art of reconstructing a vocal line from a number of takes — is always better than Auto-Tune on more naturalistic or intimate recordings, and it has recently become something a fine art. Using techniques derived from our classical music editing, we've been using Sequoia for quite a while now to provide pretty complicated vocal comping: at one point on a recent project we constructed a single word ('beautiful') and the preceding intake of breath from five different takes (yes, five — it was the way the singer sang the word itself as though it had four syllables...). Obviously, such super-fine editing requires technology specifically designed for the task, such as SADiE or Sequoia, but I was keen to see just how much could be achieved in this case with Samplitude's editing facilities, especially using its dedicated crossfade editor. In addition, on many such recordings the vocal line is usually very lightly compressed after editing, and — as a result — probably requires de-essing, and so I was keen to see what Samplitude could offer in this respect too.
The answer to the editing question turned out to be: quite a bit. The crossfade editor window (overleaf) does not, like that of the more expensive Sequoia, display the waveform, and nor can the fades be directly manipulated by mouse, but as the editing window stays on top of the main window, at least the changes made to shape and duration of fades can be seen being updated in real time. For comping at the level of whole phrases, or even separate words, Samplitude's editing capabilities were entirely adequate. It was rather slower, of course, than a dedicated editor, and it would probably take a lot more experimentation in inexperienced hands, but within these limitations we found that it could be used to produce fully professional results. On a number of occasions, though, under pressure of time, and with more complex edits to be made, I loaded the files into Sequoia to let Sequoia do what it does best.
For compression duties I turned first to the new Am-phibia plug-in, which is described as being an 'optical' compressor with very gentle characteristics, making it 'ideal for vocals'. But in use, I found that although it could certainly be gentle, it was neither colourless nor absolutely controllable, and although I can well imagine it fitting more robust vocals, it was not suitable for this project. So I tried instead the other Analogue Modelling Suite compressor, Am-track. In VCA mode, with adjustable threshold setting, this seemed much more promising, and in the end, using its pretty unique 'mix' function, we found that moderate compression mixed in at a fairly low level was just what we needed. I'm not sure that this mix function is accurately described as providing 'parallel' compression, which normally refers to a process in which an entirely unprocessed signal is mixed with a time-aligned compressed signal with a variable make-up gain, but it's a useful tool, especially when used exceedingly sparingly as it was here.
The gentleness of the compression stage meant that few artificial sibilants were created by the process itself, but as the singer has a tendency to natural sibilance, we decided to bring the new De-esser into the signal path. This is a very simple affair, with just two user-adjustable parameters — the 'tune' or frequency setting (from 4 to 12 kHz) and the amount of reduction (up to 36dB). There is also a 'listen' button which allows you to monitor what has been taken out of the signal. Sadly, I found this rather lacking for the needs of this project. It has no user-adjustable threshold, and so although it might offer a good quick fix (and with some surprisingly successful non-standard uses — see below) it could not meet our requirement with such an exposed vocal line for an entirely transparent reduction of sibilance. De-essing is, in fact, incredibly difficult to do well without introducing distortion or 'rhythmic' artifacts, which is probably why units that do work so amazingly well (such as TC Electronic's System 6000 de-esser, which we were trying out at the time, or the Weiss DS1 Mk2) are also so amazingly expensive.
We also used another Samplitude plug-in on one track. This rather weird song featured solo voice and an upright piano that was, deliberately, slightly out of tune. We comped and compressed the verse vocals, but then the producer asked if we could add a little more 'interest' in the choruses. So, with intentional punning but sonic considerations paramount (honest) we tried another new Samplitude plug-in: the Corvex Chorus/Flanger.
At first we set this up as a subtle chorus effect on an Aux buss returning on its own mixer channel, so leaving the main vocal untouched. But we found that we got a much more usable sound if we cloned the vocal track at that point (a very simple drag-and-drop process in Samplitude) and used Corvex as an insert on this new channel, processing the whole of that channel and then mixing it in, almost inaudibly, under the main vocal at the required moments. The effect was shifting and whispering, and slightly unsettling, but entirely suitable for the music. I'm not sure why that route through the plug-in worked better than standard aux buss processing, but the producer was happier with the result, and so were we.
Quiet Focus are a drummer-less jazz trio whose music features the fragile trumpet of Toby Mak, Ken Rose's classic smooth, complex jazz guitar, and the muscular acoustic bass of Pete Scherr. Each is a formidable player in their own right, and have recorded a number of CDs under their own names, but on this, their first CD together, they explore the textural and timbral interplay of their instruments as much as their harmonic possibilities. All of this was captured very effectively at 24/96, and hence was music that could very easily be diminished by heavy-handed processing and presented a unique challenge for mixing.
The CD was recorded live in Pete's studio in Hong Kong, with all of the musicians playing together in the same medium-sized room. They were positioned so that they could maintain eye contact, and the recording was made using a technique which deliberately exploited the bleeding of the instruments into each other's microphones to create the illusion of a greater space than was actually there. The requirements of mixing then were: to bring out the best of each instrument without altering its tone and timbre; to create a natural, essentially static, mix which allowed the musicians themselves to control the performance dynamics (no riding of the faders or pan-pots to highlight the soloist); and to maintain and make more uniform the sense of acoustic space.
We started with the bass. Pete has a huge technique and his playing gives both rhythm and heft, as well as a harmonic underpinning, to the music. So although we wanted to lessen a slight tubbiness at the low end (and, as a matter of course, to EQ it down where it had bled too much, as basses tend to, in the other instruments' mics) we didn't want to lose too much weight, and we also wanted to give a slightly better snap to the rhythmic contribution from the fingerboard. We used Samplitude's standard four-band parametric channel EQ and found very little to complain about in its performance: its graphic presentation made it very easy for us to see what we were doing as we swept through the bands looking for the required frequencies (using the mouse to drag directly on the EQ graphics), then as we narrowed in on them, and then finally as we dipped and boosted to taste. Remembering the 'yin and yang' of equalisation (that bands tent to interact) we found that gentle (a Q setting of 0.6) minimal addition at around 2.5kHz and slightly sharper (Q at 0.8) but still minimal subtraction at around 80Hz did the trick.
Ken's guitar, an amplified semi-acoustic, was, in principle, even more simple to deal with, as he has taken great pains to perfect his sound and wanted it in the mix just as he heard it in the room. In practice, things are never so simple, and there were certain clashing frequencies between the lower range of the guitar and the middle/upper bass that had to be tamed. Again, some minimal channel EQ did the trick, and we also brought the Am-track compressor section back into use: the compression on Ken's amplifier got him his sound, but very gentle compression helped that sound to 'sit' more easily with the others in the mix.
Toby's sound was initially more of a challenge as the trumpet is naturally much more dynamically varied than the other instruments, with a timbre that changes with volume: quietly mellow, but much sharper when louder. Of course, Toby was completely in control of these features and used them for deliberate musical effect, but there were still places where the mics were probably a little too close and so the very top edge of the note seemed a little too much, or out of place. Simple EQ did not produce the results we wanted, as no matter how little we dipped the offending frequency, it adversely affected the tone overall. Standard compression was out of the question — it would have destroyed the hallmark fragility of his sound which depends on the micro-dynamics of Toby's playing — and long hours with the Multi-band Compressor plug-in were very frustrating. And then inspiration came as we tried again the same De-esser plug-in we'd found unsuitable for the female vocals. It worked really well: set at about 9kHz, much higher in the frequency range than we'd previously thought was suitable (we adjusted it using the 'listen' facility) and with just a few dB of reduction, it gave us just what we wanted. The effect was pretty subtle: not really noticeable on an instant A/B comparison, but just enough for the listener not to register a slight 'over-insistence' in the trumpet.
Of course the instruments were mixed differently on each track, but with experimentation around these basic parameters, so the final task at track level (and ultimately at CD level: this feature did not vary very much across the different tracks) was to tie the instruments together more. The idea was not to create an acoustic space, such as Samplitude's Room Simulator is able to do so well, but to keep and make more uniform the one that had already been captured. I have to admit that I turned to the new Variverb Pro plug-in (see the screenshot on the first page of this review) for this task with some scepticism: it is exceedingly hard to find artificial reverbs that do not sound unnatural, and maybe because all of our recording, and much of our post-production work, deals with more naturalistic projects I actually have a bit of a prejudice against most of them. However, I was really pleased by the contribution Variverb made to the final mix of this project. Using the 'Club' preset as a starting point, we tweaked and tuned the early reflections and the damping until we had something roughly the same 'size' as the space captured in the recording, but then added a very slightly longer decay time. We set this up on an Aux buss and sent it varying amounts from the different instruments — comparatively more of the trumpet, rather less of the bass. The overall effect, when this was mixed back in at a pretty low level, had exactly the 'glueing' effect we were after, even if it was not entirely natural-sounding. I'm not entirely convinced that it would have worked so well with a drier recording, where there would have been a greater proportion of the reverb in the final sound, but in this instance it certainly suited the performance and mood of the music.
I'm not really sure how you can go wrong with this new edition of Samplitude. As it is, in the box, and as it says on the box, it will enable you to record, edit and mix your music to a professional level. As a recording machine it is probably as good as they get, and although no one of the new plug-ins or standard processors represents the state of the art, the combination of so many which are so good for such a relatively modest outlay makes this software a very serious contender.