Digital sampling is all about storing sounds in computer memory and using software to manipulate them - so what could be more logical than using your computer as a sampler?

Affordable samplers have come a long way from the humble Akai S612 and Ensoniq Mirage. These days state-of-the-art units such as the Roland S770 and Akai S1000 offer CD-quality sound, digital signal processing, seemingly endless sample time and so on. However, there is still a place in the market for the budget sampler - Cheetah's SX16 is a perfect example of a company responding to public demand and there is a thriving secondhand market in "yesterday's" samplers.

One alternative approach is to run a sampler on a computer. And as there's little to match the Atari ST for cheapness and popularity within musicians' circles, the door is wide open for a well-researched Atari-based sampler. With their eight-bit ST Replay (not really even a semi-pro unit) selling in excess of 15,000 units, Audio Visual Research have launched the Pro-Series 12 into the gap...

Pro-Series 12 consists of three programs and a hardware interface. The latter plugs into the ST cartridge port but derives its power from a separate PSU. The audio input and output sockets are phonos and the only other control is an input sensitivity trimpot. The hardware incorporates anti-aliasing filters on both input and output, and utilises a 12-bit analogue-to-digital converter on input and a 14-bit D/A version on the output.

To best utilise the computer RAM, the suite of three programs deal with different aspects of the sampler's operation. Editor handles the sampling, editing and processing, Drumbeat gives you the facilities of a sampling drum machine while Midiplay gives you the ability to control samples via a MIDI keyboard.

RECORDING

THE ACTUAL PROGRAM size of the Editor is small, resulting in nearly 12 seconds of sampling space in a standard 1 Megabyte Atari when sampling at 32kHz. This increases to just over one minute with 4Meg of memory. AVR have kindly used a fast-loading disk format so boot-up time is short.

The screen layout is clear, with the main sample display at the top of the screen. Beneath this are two smaller windows: to the right is the Oscilloscope, which monitors the input and shows you whether the incoming level is too low or high. Adjustments are made either from the sound source or the input sensitivity control on the hardware interface. To the left is the Spectrum Analyser, which also scans the input and shows the frequency content between 200Hz and 6kHz. This gives a rough idea of the sampling rate you should use for recording - the greater the high-frequency content, the higher the sampling frequency (although 6kHz is a little low for such an analyser). Only problem with either of these small windows is that if you drop the cursor into them, the graphics break up, but they reform within a few seconds.

There are eight options for the sampling rate of a recording: 5.5kHz, 8kHz, 11kHz, 16kHz, 22kHz (default), 32kHz, 44kHz and 48kHz. All eight options can be used if monitoring is via the interface output while only the first six are available from the Atari monitor speaker.

Having connected a sound source to the input phono, clicking on the Hear button lets you monitor that source. A useful facility is that you hear the sound via filters which synthesise the sampling rate. If you have selected 5.5kHz, your sample is likely to sound like something out of Star Trek, but as you increase the sampling rate, the fidelity improves. The payoff is a reduction in sampling time. The audio bandwidth is a little less than half of the sampling rate, although there is a hardware limitation of 12kHz. Still, the sound quality is acceptable at 22kHz and pretty accurate at 32kHz - it will take a good pair of ears to tell the difference from the original at the latter rate. The input level has to be high otherwise any noise at the input becomes over-emphasised.

At the bottom of the sample display window, the available time for sampling is displayed along with the current sampling rate. You can either use all available sample time for a sample or else type in a sample size which can either be in bytes, Kbytes (K) or in seconds (S). You can start a recording by clicking on Sample, which can be used in conjunction with the Trigger option. This requires the input to pass a threshold, given as a percentage of the maximum input level. Using the sample button has the disadvantage of not allowing you to hear the sound before you start recording. To get around this, AVR allow you to go straight from Hear to Sample by pressing the S key on the ST. The alternative is to use the Pre-sample option which continuously samples the input and stops only when you press the escape key on the computer.

At this point, the sample is drawn in the sample display window and can immediately be played forwards or backwards by clicking on the relevant button. The start and end of the sample are indicated by solid, vertical lines, while if you stop a sample in mid-play, a dashed, vertical line appears as a "ghost" cursor to mark the position. The single arrows underneath the display window are used to move the relevant cursor while the double arrows move the sample in either direction by 10% of the current width of the sample display. Zoom relocates the upper and lower cursor to either side of the sample window and you can always get back to the original cursor positions by using Reset. The Loop facility only replays the sample from start to end repetitively - proper looping occurs within the MIDI option (more later).

Beneath the Spectrum Analyser is the Paste Buffer control panel. Allowing for sufficient computer memory, Store places the portion of the sample between the two cursors into a buffer so that you can edit without destroying the original. Insert places the contents of the buffer at the lower cursor position while shifting the rest of the sample to the right. There are going to be situations where you will have part of a sample saved in the buffer but need to copy part of the screen sample elsewhere within the sample display. To help in this, Repeat pastes the portion of the sample between the two cursors to the point after the upper curser. Gap creates a space between the cursors by moving the sample beyond the upper cursor, Paste overwrites the sample window with the contents of the buffer, and Cut erases the portion between the two cursors. Finally, Swap switches the positions of the upper and ghost cursors. The bottom of the panel shows the current paste buffer size and whether it is full (F) or empty (E). You should be able to carry out most popular "cut and shut" jobs to samples with these functions.

The main editing facilities are contained in the panel to the left of the paste buffer. Most of these are useful - Fade In/out creates a fading envelope to the part of the sample between the two cursors, Volume increases or decreases the amplitude by about 12% per use, and Reverse creates a backwards sample. Fine gives you a magnified screen around either the start or end point and lets you home in on the precise position for the cursors. Shrink halves the bandwidth - if you feel that the 22kHz option is acceptable, you end up with a better result by sampling at 44kHz and then shrinking the sample, due to the way that the anti-aliasing filters work.

The final option here is Join. One of the main reasons why looping is such a difficult task is the inability to see the end point butted up to the start. The Join function converts the sample display into a split screen, with the area immediately before the upper cursor to the left and the area after the lower cursor to the right. You can then micro adjust the position of the cursors, although it is the actual sample that moves, allowing you to see the waveform at the loop point. This is similar to the method used on most visual editors for samplers, but the speed of movement of the sample on-screen is impressive. The other facility here is to Screen Draw - if you can't find a perfect loop point, you can alter the waveform slightly. It's a shame that AVR haven't implemented cross-fade looping for smoother loops but at least the Join facility gives you a visual edge in the Battle for the Perfect Loop.

Pro-Series 12 can name and hold up to a maximum of ten samples in memory, which can be recalled by pressing the Atari function keys. This brings us to the MIDI option which allows you to loop a sample in the conventional way - with loop start and end points not necessarily being the upper and lower cursor. Any of the samples stored in memory can be recalled, renamed and have a loop set. You can also assign a MIDI note to each sample, a global MIDI channel and then play back either from a connected MIDI keyboard or via the Test key. If you use a keyboard there are two modes: Trigger plays the sample assigned to the pressed key, while Keyboard plays the current sample at the pitch of the incoming MIDI note. You only get a single loop, and no alternating loop for string samples and the like. Also, there is no autoloop algorithm which you tend to find on many hardware samplers.

"Pro-Series 12 isn't intended to be regarded as a professional unit - yet it has facilities which put certain 'pro' samplers to shame."

The Pro-Series' MIDI option is intended to help in the editing of samples, which can be saved with their MIDI note and loop points in AVR's custom format. For performance aspects, you would use the Midiplay program.

Samples assigned to the ST's Function keys can be saved as a "set" of samples with their individual MIDI notes, loop points and Function key assignment.

SPECIAL FX

CREATING EFFECTS ALGORITHMS is a tricky business. This is especially true when it comes to reverb, although modulation effects such as delay, chorus or flange are easier to simulate.

Pro-Series 12 has its own range of effects, some of which are primarily presets. The Hall and Room reverbs are more like multiple echoes, but quite usable. Flange, with editable speed and depth, is er, effective, while Ramp turns out to be a vibrato simulation with control over the speed. Other effects have depth and volume control; Echo and Reverb are two such effects, while Shift increments the pitch of each echo. Finally, Multi has two sets of depth and volume controls, one of which is for the feedback loop.

The effects can be used to alter a sample in memory or to act directly on the sound at the input in real-time. The idea of using the software as an effects processor is rather interesting but has the drawback that you can't alter any of the settings while listening to the effect. I wouldn't go as far as to say that if you're not using the editor for working with a sample you could use it as an extra effects unit, but the flange effect is certainly good enough to be used on, say, vocals on a demo. It all depends on what else you have available.

FILTERING

FILTERING IS ONE area where software samplers often score over their hardware counterparts. Sampling often generates extraneous noise which can be cleaned up by the removal of unwanted high (hiss) and low (rumble) frequencies. Alternatively you can use filtering to create special effects, like reducing the 200Hz frequency content to make a sample sound like it's coming from a transistor radio.

Pro-Series 12 gives you two filtering options: Slow and Fast. Slow offers low-pass, high-pass, band-pass and notch filters with two frequencies being available for the last two to set the range over which the filters work. Fast gives you a wider selection, with the above four and additionally boost, no dc, bass and treble. The difference is that you only have control over the centre frequency where a range can be used, the rest of the parameters being preset. Low- and high-pass also have a different curve gradient. Slow gives you more accurate filtering than Fast, but takes longer to achieve results. In fact, notching a frequency on a ten-second sample using the slow filters took over ten minutes. Needless to say, this is something we want to get right first time. To this end, there is the Response option which draws a graph showing the effect that the chosen filter will have, and the FFT - Fast Fourier Transform. This draws a three-dimensional graph showing the frequency content against time for either the entire sample or the portion between the cursors. Unfortunately, there is no vertical axis or ability to zoom in but the graph shows approximately where a nuisance frequency exists.

The only other problem is that once you set the filtering into motion, there is no way to abort apart from turning the computer off.

CHANGE OF PROGRAM

HAVING EDITED SAMPLES and saved them via the Editor, you can load them into Midiplay for playing via a MIDI keyboard. This program holds up to 128 samples in the ST's memory and uses four-note polyphony, with the option of running four voices monophonically. Samples are mapped to groups of keys on a MIDI keyboard, up to four maps can be stored in memory along with the ability to switch between them. Volume and tuning can be set for individual samples but the entire map can only be assigned to a single MIDI channel. If you find that the MIDI loop points set within the Editor are incorrect, you can re-edit those here. Polyphonic playback is of a lower quality than monophonic, but that is to be expected. Very basic, but adequate.

Drumbeat is effectively a sampling drum machine. Only 22kHz samples can be used, so sampling at 44kHz and "shrinking" is the order of the day. Up to 15 samples can be used within the 50 patterns and you can have a maximum of 100 steps per song. The pattern layout is clear and easy to use (a bit similar to the MIDIDrummer ST program). Each sample can have a MIDI note, channel and velocity setting which can be saved as a "kit", and the numeric keypad to the right of the ST can be used to play and record the samples in real time. MIDI Start, Stop and Clock are supported, and you can trigger samples from an external MIDI source (keyboard or MIDI pads) - three different samples can be triggered by a note, depending on the velocity. The polyphony is again restricted to four notes.

VERDICT

IN SPITE OF its name, Pro-Series 12 is not really intended to be regarded as a professional unit. It has, however, facilities which put certain "pro" hardware samplers to shame - the real-time effects and digital filtering, for example. Sound quality is good - certainly on a par with most cheap samplers if used for a single sample.

For someone who already has an Atari ST and simply wants to delve into making good-quality samples, Pro-Series 12 is tailor-made. Yes, there are cheaper, eight-bit samplers, but their lack of quality limits their potential uses. When used in tandem with Drumbeat, you have the makings of a simple, but powerful, drum machine. It's a shame that you're limited to four voice polyphony, but you can't have it all...

Whether any semi-professionals would consider using a 4Meg ST with Pro-Series 12 purely for spinning in long vocal sections is doubtful, but look at it this way: a system bought specifically for this purpose would cost around £800; where can you buy a sampler which will give you 60 seconds of sampling at 32kHz for a similar amount?

Price £245 including VAT and p&p.

More From Audio Visual Research, (Contact Details).