Creamware Powersampler DSP

It would take a lot to convince your average engineer, producer orprogrammer to abandon their trusty hardware sampler, but Creamware may finally tip the balance with their new PCI sampling card. Martin Walker never says 'never' again...

The user interfaces of most professional samplers can already be moved onto desktop computers, taking advantage of the increased display and control options that a monitor, mouse and keyboard can provide. Some manufacturers have even set up file‑exchange options between computers and their samplers, in order to allow the user to share storage hardware between the two machines. Creamware's new Powersampler, however, allows complete integration between dedicated sampling hardware and your computer by packing a complete hardware sampler onto a PCI expansion card, which can be installed into a Mac or PC computer. This therefore allows the sampling hardware to access your computer's RAM, vast hard disk space, and a large‑screen editing environment, while leaving the computer's CPU untaxed. Despite the 32 phase‑locked stereo voices, 16‑channel multitimbrality, 24‑channel DSP mixer and effects processing which the Powersampler offers, the dedicated SHARC DSP chips impose virtually no overhead on your main processor

Of course, computer‑based sampling isn't new — we've had Soundblaster cards with SoundFont sampling for some years, for example — but professional musicians are not willing to accept the noise levels and budget A‑D/D‑A conversion of such systems. The SB Live! and Emu APS cards were both a step in the right direction, but neither has felt like a traditional sampler in use, and their fixed 48kHz sampling rates have caused problems for anyone wanting to transfer files digitally. However, the Powersampler is, without doubt, a professionally specified card: the 24‑bit/96kHz sampling capability, 32‑bit internal processing, and an ultra‑low latency (under 2mS) which it provides surpass the technical specs of a number of stand‑alone hardware samplers. In addition, in order that the card can act as the I/O hardware for a variety of popular recording packages, a generous range of drivers is available, including ASIO, EASI, MME, OMS, and Nemesys (for their Gigasampler).

You can use a second card to double the number of voices to 64, and for those who need lots of physical I/O, an optional breakout box will be available shortly, containing eight further analogue ins and outs, all with 24/96 capability. Existing users of the Creamware Pulsar system (see SOS May '99) can also use Powersampler as an expander card. Alternatively, you can buy a special software version for loading directly into Pulsar or SCOPE as another device (see the 'Pulsar/SCOPE Connection' box on page 80).

Hardware & Installation

The card at the heart of the Powersampler system.

The Powersampler hardware consists of a 6.5‑inch‑long PCI expansion card, which should fit happily into any PC or Mac. It houses three 32‑bit SHARC DSP chips (identical to those used in other Creamware products, such as the Pulsar), several Quicklogic chips, and one of AKM's AK4524 codec chips for A‑D and D‑A duties. The backplate houses a pair of quarter‑inch stereo jack sockets providing a single stereo input and output operating nominally at ‑10dBV, a pair of six‑pin sockets for MIDI In and Out, an S‑Link socket for the optional 8‑in/8‑out breakout box, and an eighth‑inch mini‑jack for S/PDIF In/Out. This last connector does seem an odd choice, but has the advantage of only using a single socket to service both digital In and Out — a stereo 3.5mm jack to twin phono adaptor lead (not supplied) will interface neatly to most DAT or other digital recorders. A pair of of six‑inch adaptor cables with 5‑pin DIN in‑line sockets is included for MIDI I/O, suitable for attaching any standard MIDI leads.

There are a further two connectors on the card itself. The larger of the two is a multi‑pin expansion card connector (there are no details on the plans for this yet), while the other is an S/TDM SCOPE buss connector, capable of carrying up to 128 32‑bit data channels, which you can use to link with a second Powersampler card. Also bundled with the hardware is a CD‑ROM containing the Powersampler's software, drivers and user manual, and a second CD‑ROM containing the same 400Mb Ultimate Sample Collection included with the Pulsar system.

The review model was supplied with PC software only, but Creamware have told me that the Mac version is already at the beta‑testing stage. I installed the hardware in my Pentium II 450MHz PC with 128Mb of RAM. After seating the card firmly in one of its PCI expansion slots, the hardware was correctly detected by Plug and Play during reboot, and the drivers installed automatically. The software is then installed separately by running the 'setup.exe' file on the main CD‑ROM, and serial‑number protection is used. The installation went smoothly on my machine, adding 38Mb of files to my hard drive.

Software Overview

The Toolbar Environment window lets you choose from a wide variety of driver types, depending on which software you use.

It took about 25 seconds on my PC for all the appropriate DSP code to be downloaded to the SHARC chips, following which the small Toolbar window appeared. From here, you can access the two main components of the system, the STS 3000 sampler and the 24 Channel Mixer. The Toolbar also has drop‑down menus for Environment and System settings, and launchpads for the five larger windows from which the Powersampler is controlled (shown on page 78): the Main Window for the STS 3000 sampler, the KeyGroup List, the Sample Editor, the 24 Channel Mixer, and the File Browser. All windows have a mock brushed‑aluminium finish, and each has its own Close button, along with a 'Stay On Top' button that gives it a higher display priority than other windows.

The Toolbar window's System options include whether the board acts as master using its internal digital clock, or as slave using an external clock from the S/PDIF input. In Master mode you select a suitable sample rate from 32kHz, 44.1kHz, 48kHz, or 96kHz, and the current rate is displayed. You can also adjust latency — there are five buffer settings whose latency values vary depending on sample rate. At 44.1kHz these are 25mS, 13mS, 7mS, 4mS, and 3mS, and these values are halved when running at 96kHz. However, the 96kHz setting uses twice as much DSP power, so while you can still use the card as a 96kHz soundcard, the sampler is deactivated unless you select 48kHz or lower sampling rates.

The Environment menu allows you to launch a further Settings window, from where you can adjust which drivers are available in the Mixer and within other Windows applications. The available drivers are arranged in three groups: the right‑hand column contains a number of Wave (MME) options. The simplest is '16‑bit Stereo Wave', which is the only driver that doesn't need extra DSP power to function — all the others reduce maximum voice count slightly. There are three other Wave options (Multiple 16‑bit Stereo Wave, Multiple 24‑bit Stereo Wave, and Multiple 24‑bit Interleaved Wave), and these can be activated if your sequencer application doesn't support ASIO or EASI drivers. They provide eight stereo WAV inputs and outputs, and reduce the voice count by between two and four voices. Windows must be restarted after any changes are made in this section. Inside applications, the MME drivers simply appear as Powersampler Play 1‑8, and Powersampler Rec 1‑8.

The majority of users will be able to choose one of the three options in the left‑hand column — EASI, ASIO 16‑bit, or ASIO 24‑bit, depending on whether they use Emagic's Logic Audio or Steinberg's Cubase VST. All three options reduce the total voice count by between three and six voices. You can also simultaneously activate any combination of the DirectSound, TripleDAT, and Gigasampler drivers; Tripledat and Gigasampler drivers each reduce voice count by an additional two to four voices. Creamware stress that you should only activate the minimum combination of drivers, to make the most of your DSP power.

From this Settings window you can also select whether or not the sampler is loaded, boost the input sensitivity of the analogue inputs by 12dB, to cope with low‑level signals, and set up MIDI Routing. There are two choices for the latter — External connects the sampler directly to the MIDI In socket with 1‑2mS latency (for using it as a stand‑alone device), and Internal connects the MIDI In to your sequencer, and the output of the sequencer to the sampler.

24 Channel Mixer

The Main Panel shows the current Multis, along with their many parameters and overall RAM consumption. Further parameters can be accessed through the sub‑window.

The mixer window is used as a 'junction box' to route signals from the hardware inputs or software driver outputs to the hardware outputs or software driver inputs. Up to 24 channels are available, depending on the combination of drivers you have selected. The default background environment configures the mixer with just four channels — Wave L and Wave R for the 16‑bit Stereo Wave driver, alongside Anlg L and Anlg R for the two analogue input sockets. Three additional ready‑configured environments are also included. 'Sampler' replaces Wave L/R with eight channels comprising the sampler's main stereo output and six individual mono outputs, while Anlg L and R are routed to the sampler L and R inputs for recording your own samples. There are also two record monitor outputs, and a further two from the sample editor, so that you can hear the raw sample. The other two preconfigured environments set up the mixer for use with Cubase and ASIO drivers, or with Logic with EASI drivers. In both cases the sampling channels are as before, but channels 17 to 24 carry the signals from the eight ASIO or EASI outputs respectively.

Each of the 24 channels is identical, with an input routing box, delay and chorus level controls, a pan knob, Mute and Solo buttons, pre‑fader clipping indicator, a Mix button to add the channel output to the stereo master buss output, volume fader with post‑fader meter, Stereo Link button, and output routing box. The Mixer displays eight channels side by side, and four page switches along the top of the mixer let you choose between viewing channels 1‑8, 9‑16, 17‑24, and Effects. This last option displays the master controls for the stereo delay and stereo chorus. The delay has individual channel delay times and variable feedback with high‑frequency damping and cross‑feedback switch, while the chorus has delay, rate and depth controls, along with variable phase between the two channels. Both effects have output‑level meters, clipping indicators, and output routing selection, and although they are fairly basic to save DSP power, they're still useful and effective. The stereo Master section is always visible, and contains master level controls for the effect sends and returns along with a master fader, twin clip‑indicating level meters, and output routing boxes.

Input and Output routing selections are made by right‑clicking on the appropriate boxes — there is a host of available options, grouped into categories: Mixer (all 24 possible channels plus the main, chorus and delay outputs), Analogue sockets, Digital sockets, Wave I/O (for use in MME applications) and Sampler I/O. Outputs can be sent to multiple destinations, although confusingly only the most recent assignment appears in the text box. You can type more meaningful names for each channel in the boxes provided, and if you have carefully set up a mix then you can save it as a preset that can be reloaded into the same environment later on. Given the number of routing options, the mixer can initially be confusing. However, at least invalid options such as connecting an output to itself are prevented, although the resulting error messages (a highlight of which includes "connection Add 32.Out to Mixer Master Gain.In generates recursion without delays!!!") could do with better translation. Thankfully the on‑line electronic manual is quite readable, and does guide you through various practical examples, including use with Cubase and connection of external hardware such as DAT recorders and digital effects.

Basic Operation

The 24 channels of the Mixer are used to create a stereo monitor mix, and to route the various audio and sampler outputs to their destinations.

Anyone who has used an Akai sampler in the past will find the Powersampler very familiar in many ways, even down to which parameters appear in which screen display. The highest level is 'Multi', containing up to 16 Programs, with each one assigned to any of the 16 MIDI channels. Each Program contains one or more KeyGroups, and each one of which can contain up to four sample Zones, allowing you to load stereo samples or create up to four velocity‑switched ranges. The settings in a Multi can be saved as a Preset, which lets you use the same samples in various songs without having multiple copies of them scattered about, although this doesn't contain the samples used. Further versatility is provided by the Program Pool, which lets you reference up to 999 Programs stored across various folders and drives for easy access. You can create multiple pool Presets, and the first 128 positions in each pool can be called by MIDI program changes. You can either load Programs in the pool into RAM for quick access, or you can load them on demand. Pools are an ideal way to access the many sounds included in GM SoundFonts, for instance.

The upper panel of the STS 3000 window displays the list of Programs within the current Multi, alongside individual Mute and Solo buttons. Three views are available, each of which offers additional options: the Main page displays MIDI channel, level, and pan values; the Additional page allows you to play the sample through the individual mono outputs rather than the main stereo one, and to determine its level, transposition, low and high keyboard range, and priority; and the Memory page displays each program's Sampler Memory (as a percentage of the total memory allocated to the program), Total Memory (as a percentage of available RAM), and Size. The low‑level parameters for each Program are contained in a separate sub‑window that drops down from the Main Panel. There are eight possible views within this window, from which you can alter settings relating to Loudness and Pan modulation, Filter modulation, LFO1, LFO2, Pitch modulation, MIDI, Tuning and the Soft pedal.

The KeyGroup List window can be launched from both the Toolbar and STS 3000 windows, and allows you to set up samples across the keyboard. Their relative tuning, loudness and pan position can be adjusted, and settings are also available here for a 12dB/octave resonant low‑pass filter and two envelopes — as with Akai samplers, Env1 normally controls amplitude, while Env2 defaults to other duties such as filter, pitch, and pan modulation.

Loading, Recording & Editing Samples

The Sample Editor has a resizable waveform window for detailed editing, as well as a split‑display mode to help when finding loops.

If you already have a sample library on Akai CD‑ROMs, using the Powersampler is simply a matter of opening the File Browser window from the Toolbar, then dragging some suitable programs across into any of the 16 slots available in the STS 3000 window. If you are using a CD‑ROM, all Akai partitions are read and the entire contents alphabetically sorted. If you use the Show Partition option the display will end up sorted in a similar order to a CD‑ROM contents listing, while Hide Partition leaves only the folder names in order to facilitate finding particular sounds. You can choose to view programs, samples, or both. It might be useful, though, if Creamware added an option to display file sizes in a future update!

When you load an Akai program, all of the original program parameters are retained, including its MIDI channel, filter settings, level, pan, and so on. All of the sounds I loaded sounded exactly the same as they had in my Akai sampler. I've used several Akai‑loading CD‑ROM utilities on the PC, but this one is by far the most foolproof and elegant. Loading in SoundFonts is almost identical — you drag them into an appropriate slot from a CD‑ROM or hard drive — but if they contain a number of presets, you can only access the ones beyond the first by dragging the SoundFont into the Program Pool, which lists all the presets contained within it, allowing you, theoretically, to drag any of these into the Powersampler slots — however, the application crashed every time I tried this. You can save programs individually, or with their associated samples as a volume, and you can also Export volumes in STS, Akai S1000 or Akai S3000 format into a single folder for copying onto CD or removable disk, should you wish to transfer them to another system.

Samples in Akai, WAV, or AIFF formats can be dragged directly into any Zone in the KeyGroup List window, although they must be either 8 or 16‑bit, mono or stereo. New mono or stereo samples can be recorded as well, using the Sample Editor window. Recording can be triggered manually, by a MIDI note, or by audio threshold value, and a useful Pre‑Record feature captures a user‑defined period before the threshold is reached, to make sure that transients are always captured. A comprehensive tutorial on recording your own samples is included in the electronic manual.

The Sample Editor window's main display shows the waveform of a selected sample, which can be resized to be as large as you like (hooray!). You can zoom in and out of any part of the waveform with single‑sample accuracy, perform all the basic operations such as cut, copy, paste, delete, DC offset removal and normalisation, and audition your sample complete with loop if enabled. The display can be split into two halves to view loops more easily, and although there are no crossfade looping options you could easily launch an external sample editor to do this. Only mono samples are currently supported, although a stereo editor is apparently in preparation.

In Use

This is the way I set up Cubase to add real‑time software effects to each of the Powersampler outputs. The deactivated inputs are for a further eight channels of normal audio recording.

I found working with Powersampler very similar to using my Akai hardware sampler, except that I never had to leave the comfort of my PC screen. The interface looks gorgeous, providing a professional yet familiar set of screen displays, and the beauty of the multiple‑window design is that you can leave the Toolbar permanently on screen along with your sequencer, and then open as few or as many of its windows as you need at any time. My only disappointment was the frustrating sluggishness of some operations, such as selecting waveforms in the Sample Editor and dragging sounds from the file browser to the main window — the latter often resulted in sounds being dropped into the wrong slot because the position lagged behind the cursor.

Audio quality was very good compared with my own Echo Gina card, showing improvements in transient detail similar to those exhibited by the other three soundcards with AK4524 converters that I've reviewed recently: the Terratec EWS88MT, the Aardvark Direct Pro 2496, and the SEKD Siena. RMS background noise was similar to these at ‑93dB for 16‑bit and around ‑100dB at 24‑bit, but I doubt that many people will be tempted to use the 96kHz sample rates for audio recording, since you lose the facilities of the sampler. I had no problems running it with 7mS latency, and despite my reservations about the sluggish interface this doesn't affect real‑time performance, which is excellent.

Final Thoughts

The Powersampler doesn't have any direct competition at the moment, although it's an open secret that Yamaha are also working on a PCI sampler card, but entry‑level hardware samplers are also being aggressively priced. Akai's S2000 is almost the same price with its full complement of 32Mb of RAM and 32‑voice polyphony, though increasing the outputs to eight adds another £150. If you need 64‑voice polyphony then buying two Powersamplers will cost you £900. At this price you might be better off buying a Pulsar and the STS 4000 software separately for a more powerful system, or considering the option of a 64‑voice hardware sampler such as Emu's ESI2000 for about £630 or Yamaha's A3000 for about £700 — both these stand‑alone units are expandable up to 128Mb of RAM, but only 4Mb and 2Mb respectively are supplied as standard.

However, none of these alternatives brings an easy‑to‑use professional sampler into your computer environment in the same way as the Powersampler, which really scores by being able to share your existing hard disk and CD‑ROM drives as well as however much RAM you can fit in your computer. The system also offers all the benefits of large‑screen editing and the possibility of automated effects on each output. Although it only has a single stereo In and Out you can still add effects to all eight of the sampler outputs individually, and the forthcoming breakout box will provide more hardware Ins and Outs for those who want to record live ensembles at a single sitting. The main limitation is its 32‑voice polyphony, which is rather low by today's standards — and of course with Cubase or Logic drivers enabled you end up with just 26 to 29 voices.

As is inevitable with the first release of such an ambitious product, I found a few small bugs, but these were mostly cosmetic and Creamware have already dealt with some of them. For me, the most important feature was the total integration of the sampler into my computer. Not only did this make sample editing more pleasurable, it made the dedicated Syquest and CD‑ROM drives I currently use with my Akai sampler instantly redundant. Being able to store everything relating to a song in one place was truly liberating.

Creamware have a strong product on their hands with the Powersampler. It was definitely the right approach to model it on classic Akai S1000/S3000 principles — it just makes it so easy for people with existing libraries to move sideways with the minimum of fuss, and I can see many musicians who dream of integrating their sampling with their computers doing just this. With support for both Mac and PC, plus Akai and SoundFont compatibility, it has a huge potential market, and I expect it to sell really well. In fact, I'm seriously tempted to buy one myself.