Armadillo A616 Sampler
Still largely neglected by the software houses, the Acorn Archimedes is the host machine for the Armadillo sampler. Vic Lennard investigates a powerful sampling package for a powerful computer.
Powerful it is, but without the right software the Archimedes computer is not going to be welcomed into music studios. Is Armadillo's upmarket sampling package the break it's been waiting for?
THE A616 IS an eight-voice stereo sampler, comprising a 1U-high rackmount unit and software which will run on any of the Archimedes series of computers. The hardware has two unbalanced audio inputs and outputs as well as a headphone socket and a digital signal output (SPDif), along with MIDI In, Out and Thru (spelt "through" in a quaintly English fashion) ports. The unit is connected to the computer via a ribbon connector and a special Armadillo podule which plugs into the back of the Archimedes.
Sampling frequencies of 44.1kHz (CD) and 48kHz (R-DAT) are offered, and the amount of sampling time available is dependent on the memory of the computer. To give you an idea, on a 4 megabyte machine with all available memory freed for the sampler, just under 20 seconds stereo (40 seconds mono) is available at 44.1kHz.
The software comes on two disks: a program and data disk and a "key" disk which has to be inserted into the drive at the relevant moment. This may seem like overkill (how many people copy hardware?) but Armadillo are planning a program called MIDItools which will require this protection.
ANY PIECE OF software is only as good as its user interface. With this in mind Armadillo's Highnote goes a long way towards making the sampler very intuitive to use. The approach is modular, in that various modules exist on the menu bar at the bottom of the screen. Two of these belong specifically to Highnote - the Sample Editor and Voice Setup. Using the Sample Editor module, you can make samples, play them back either from an on-screen keyboard or a connected MIDI device, and edit them in a variety of different ways.
Up to eight samples (or Voices) can then be put together into a Patch using the Voice Setup module. Voices and Patches stored in the Archimedes are accessed via the Object Manager (which organises the internal memory) and can be saved to disk via the Floppy or Hard Disk modules. The mouse on the Archimedes has three buttons; the left-hand button selects an item for movement around the screen; the middle button calls up menus dependent on screen position, and the right-hand button modifies parameter values. Highnote incorporates double-clicking on the left button for selection and opening of an item in one go. Movement around the screen is by the usual click-and-drag technique.
A DOUBLE CLICK on the Sample Editor icon opens up a sample window. This represents a sample in two ways; Overview and Zoom. The small Overview graph shows the entire sample and has two green triangles underneath it. Dragging these sets up the portion of the graph which is seen in the larger Zoom graph above the Overview window - up to eight zoom-ins can be made down to individual sample points. To make a sample, a click on the area to the left of the graphs brings up the Sampler menu which takes you to the Sampler window. Here you select the sample time and type (Left, Right, Mono mix or Stereo) and whether sampling is triggered by a threshold level. This is given as a percentage value and is set via a slider which is next to the two VU meters showing the level of the incoming signal. You can also set an amount of pre-trigger time. The amount of time set is looped around until recording is started, and the signal for the allocated pre-trigger time directly before this moment is added to the beginning of the sample. This prevents fast attacks from being missed. Turn the Monitor switch on to hear the incoming signal, press Record and you have a sample. To monitor the result, press Play. One oddity here is that the level monitored on the incoming signal appears to be much higher than that heard on playback. Finally, you can Name the sample from this window.
The Zoom window now shows the right-hand and left-hand halves of the sample above one another. To hear the sample once all the way through, click on the left mouse button. To playback a portion of the sample requires the setting of markers - click with the left button in either of the graphs and a yellow line appears with a rectangular indicator above the Overview. Set a second marker in the same way. A click with the right button just to the right of the left marker inverts the colouring of the area and makes it active. Playback now plays only the highlighted section. In the Sampler menu, Play mode lets you choose whether the area is played once (Single Shot), repeatedly (Continuous) or automatically when a new active area is selected. Markers can be moved by the usual click and drag method and deleted either via a menu option or by using the dustbin.
Using the Marker menu, markers can be set to various positions like the start or end of the sample, before or after the next/last zero crossing, and to the previous position. Markers can also be converted to loop start/end markers, in which case the indicator changes shape to a triangle. This has no effect when playing back a section of sample, but is important under the Voice Setup module.
When moving markers and zooming in and out, the windows can take a fair time to redraw. The resolution of the graph can be changed by going to the configure option for the Sample Editor module. By changing Fast Plot to "On", not every sample point is redrawn, speeding up the redraw process. Other options here include the default play mode, marker sensitivity (for setting how close to a marker you need to be to pick it up) and whether a horizontal line is drawn across the centre of the sample. These choices can be saved to disk for booting up with.
ONCE AN AREA is selected, you can either edit both halves of the stereo together by selecting the Edit menu from the Overview, or edit each half separately by getting the menu from the left or right graph. There's every chance your sample hasn't been recorded as loudly as it could have been, so Gain lets you "normalise" it - take the sample point with the highest amplitude to maximum and increase all other points proportionately - or increase/decrease the amplitude in steps of 3dB. The sample can be Reversed, giving interesting effects when only one of the stereo pair is edited, or set to Zero to create a blank portion. Fade lets you set a linear fade in/out from silence or from 3dB down on the current amplitude for less severe fades. Delete removes the area permanently and closes the gap in the sample.
"Armadillo promise sequencing with SMPTE time code, sample analysis and resynthesis and real-time effects in the near future."
Chopping a sample up and moving parts around requires more than just an active area. A marker is required to show where parts are to be placed. Any marker can be made active by using the appropriate option in the Marker menu, at which point the colour changes from yellow to red (warning: this option can seriously damage your sample). Insert takes the marked area and creates enough space to copy it after the active marker. Unfortunately, multiple insertions are not available so copying an area more than once can be tedious. Overwrite performs similarly but doesn't create space after the active marker, while Move erases the area from its original position once it's been copied. Add combines the active area with the area currently following the marker (which can cause clipping), while Average does the same but halves the final amplitude (to avoid clipping). To minimise glitching when samples are cut and shut, Crossfade lets you define an area over which one part fades out as the other fades in. This includes an option for the loop points which gives crossfade looping. This technique lets you create smooth loops by taking a small portion of sample before the start loop and mixing it in with the area before the end of the loop. It is usual to be able to choose how wide an area of sample is taken before the start, which is not offered here, but the results are good. This is helped by continuously zooming in so that you can see precisely where the loop should be.
There are times when a stereo sample may need to be split into two mono sections and each side treated separately. The Split Sample option in the Sampler menu keeps one half of the sample under the current name and renames the other half automatically. To create a stereo sample, use Make Stereo from the same menu. The current mono sample inhabits the left side and the right one is then blank. Set an active marker in this blank right side and then drag an active area from another sample on screen. Care has to be taken when editing because there is no undo facility - edits are permanent and destructive. I would have preferred to see the original of a sample buffered for recall if necessary.
The Archimedes' multiple window environment allows you to view different aspects of the same sample at the same time. Unlike the Atari ST and other similar machines, a window does not have to be active to move it which makes this software very quick to use. For example, you can have three or four different samples on screen and cut out bits of each and insert them into another window to create a new sample. The way most computers work is to use a clipboard buffer because of the lack of processing power which makes the whole procedure much slower.
HAVING MADE VARIOUS samples, there are likely to be two different situations for playback. The first of these is to switch between various samples in memory and for this the Channel Grid is used. This is found in the Voice Setup menu and consists of a Voice list with a channel grid next to it. Use of the term "channel" is likely to cause confusion with MIDI channels, but in this case it refers to the eight voice channels. Samples are dragged to this window and some or all of the available eight voices may have samples assigned to them. Beside each sample is the length (in seconds), stereo image position and whether the loop created for that sample is to be on or off. Clicking on a voice brings up a menu allowing you to change the note at which the sample plays back as recorded, and detune the note in 50ths of a semitone. It also gives the start and end times of the loop. Under this menu there is access to the System Configuration which sets the Master Volume/Tune and the overall number of voice channels. This page can be found in many of the menus around screen.
The way in which Armadillo have used the four MIDI modes is rather unconventional. Mode 1 (Omni on, polyphonic) accepts all MIDI channels and plays each voice consecutively. This is fine if you've assigned four voice channels to each side of a stereo sample, but if there are eight mono samples, each with one channel assigned, they play alternately. Mode 2 (Omni on, monophonic) works from all MIDI channels but only plays the first voice. Mode 4 (Omni off, mono) gives each voice a separate MIDI channel starting from the Base Channel in the MIDI Options menu. Mode 3 is reserved for patches (more later).
Selections from the MIDI Options menu control the playback from the channel grid. This has the MIDI mode setting and various other functions, the less obvious of which follow. Legato is often taken to be the playing of a second note immediately after the first without a gap. The use here is more in keeping with that for a wind instrument. In MIDI mode 4, if a second key is played before the first is released the pitch changes but the sample doesn't restart (single triggering in analogue terminology). Sustain plays a sample through to its end - one-shot mode.
Should you not have the hardware to hand, an onscreen MIDI keyboard can be called up and have a MIDI channel assigned to it. Key numbering follows the Roland (middle C is C4) standard, with just under six octaves as the standard screen display - though you can zoom in and out. The same menu also has settings for sustain time (which delays the note off), the velocity sent by a keypress and whether sideways mouse movement controls glissando or pitchbends (not currently available).
While I was using Channel Grid mode, a variety of quirks occurred. Changing MIDI mode made the sampler forget the current setting of the stereo switch in the MIDI Options menu. Samples suddenly played back one side at a time and the stereo switch had to be turned off and then back on. Similar confusion arose when rearranging the voice channels. Samples played from the on-screen keyboard were often an octave out immediately after this change. Checking voice detune showed this to be true, and yet the tuning was restored on exiting the menu. Odd. The manual states that the channel voice allocation numbers will shift from one sample to another when a MIDI program change is received, but I couldn't get this to happen. Hopefully these problems will be cured in later software revisions.
WHILE THE CHANNEL Grid is fine for checking out samples against each other, it does not support multitimbrality. This is in the jurisdiction of the Patch List.
To create a new Patch, the Patch List window must first be opened from the Voice Setup menu. Once created, master volume and tuning parameters can be set per Patch as can the number of voice channels. As with the Channel Grid, samples are then dragged to the Patch window where each can be allocated a keyboard zone. Information such as root note, detune and MIDI channel are all shown within the Patch window, and can be changed via the individual menus for each Voice. This aspect of having duplicate items in many menus is an excellent feature of Highnote and saves you remembering where everything is.
"Software is only as good as its user interface, and Armadillo's Highnote goes a long way towards making the sampler very intuitive to use."
Setting up the high and low notes for the zone and root note is a rather ingenious process. Clicking on a voice highlights it and makes it active. Playing notes from either the on-screen keyboard or an external MIDI keyboard changes the parameters one at a time - you can even put the high and low notes in the wrong way round. There are a further four toggle switches for each Voice; Sustain Loop as before, Sustain and Velocity on/off as in the MIDI Options menu and Fix which keeps a Voice at constant pitch no matter where on the keyboard it's played. You could have a different sample on each key and all settings are kept within the Patch which then sits in the Patch List.
Channel Grid and Patch List are mutually exclusive - you can only have one of them active at a time. This is determined by the MIDI mode. Mode 3 (Omni off, poly) makes the Patch List active while any of the other three modes favours the Channel Grid. There are certain aspects which are, as yet, unavailable. The obvious of these is the inability to overlap keygroups. This means that you cannot overlay samples or use a positional crossfade or velocity switch to move or change from one sample to another. You cannot delay the sounding of a sample. For instance, it would be nice to take a piano and string sample, overlay them and then set a delay on the attack of the strings giving a piano attack and a string release. But it's not possible here. Another shortcoming is that if the polyphony is exceeded, a warning flashes on screen and all audio cuts out. It would have been a better idea to have simply cut off the first or last note played. Finally, try as I might, I could not get Patches to respond to a MIDI program change.
GENERALLY, THE A616 is easy to use and the quality of audio is excellent. I used it on some demos for a record company where seven layers of vocals in a stereo spread lasting some 18 seconds had to be laid on tape at various places - the A616 handled the job admirably. Using a full-size colour monitor and the Channel Grid option, the entire process was finished in 15 minutes with no audible difference between the originals and the samples. The four times oversampling used on playback helped give the results a particularly smooth finish.
One shortcoming did come to light, however: Highnote doesn't recognise any MIDI controller information. No pitchbend, no modulation wheel and so on. Also, there's no MIDI implementation table with the A616. The manual is reasonable but would benefit from more examples - especially where the channel grid and patches are concerned.
Sadly, although I didn't get the chance to try the A616 with the Inspiration sequencer, it wouldn't have worked as Highnote precludes the use of other Archimedes programs. Armadillo are intending to convert the necessary software routines to RISC-OS (the operating system used by the Archimedes) to allow compatibility by the end of the year.
THERE ARE BOUND to be comparisons made between this and other 16-bit samplers such as the the Akai S1000. Cost apart, what are the advantages of using a sampler based on a computer? The most obvious are the same as those applicable to hardware and computer sequencers - the screen, for example. So many editing functions need to be clear in order to be best used. Removing pieces of samples, for example, is practically impossible on a hardware sampler yet it can be carried out within consummate ease on the A616. Similarly, carrying out different edits to the two halves of a stereo sample wouldn't be practical on anything without a display. And given the same edits, the A616 will out-perform any hardware sampler due to the speed of its processing. So what are the disadvantages?
Looping is one of the most important features of any sampler and yet the ability to obtain glitch-free loops here is limited. No autolooping, no ability to check level and gradient of the two portions to be looped and the lack of any option of crossfade looping width are not limitations I would expect to see in a top-line sampler. No ADSR envelope or filtering, no release or alternating loops. More sophisticated facilities like cut-and-play lists and, of course, direct-to-hard drive recording are already available on Atari ST (Hybrid Arts Adap I/II and Digidesign Sound Tools) and Mac (Sound Tools) systems. The list of shortcomings isn't endless but is certainly too long.
Armadillo promise sequencing with SMPTE time code, sample analysis and resynthesis, real-time effects (echo, reverb, vocoding) and click/scratch/noise removal modules in the near future. More to the point, when the direct-to-hard drive hardware is available, we're likely to see this being used for digital editing - an area which has been poorly covered by reasonably-priced systems.
And the cost? An Archimedes A440 (4Meg RAM and a 50Meg hard drive as standard) with an Acorn monitor costs £2319 while the Armadillo A616 soft/hardware costs £1295, so it's not a cheap system. You could get an A3000 with 2Meg RAM (max) and a monitor for £968 but less RAM means less sampling time, and more limitations. (All prices exclude VAT).
Armadillo certainly have some exciting ideas planned for the future, but some of the more basic facilities need to find their way into the package sooner rather than later. What we have here is a bright start to a system which shows much promise. However, Armadillo have a fair way to go before the A616 is the system they intend it to be.
Price £1295 plus VAT.
Thanks to Phil Brown of Pandora Technologies for the loan of an Archimedes A440.
Review by Vic Lennard
Previous article in this issue:
Next article in this issue:
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!