Soft Touch

Sep 1, 2003 12:00 PM, By Len Sasso

The time has arrived. With a reasonably powerful laptop computer and an affordable collection of software instruments, you can easily take your studio on the road. And with suitable MIDI controllers and an audio interface, you can even take your portable studio onstage.

Software synthesizers come in a confusing array of formats, but they have a lot in common under the hood. In this column, I'll cover the basic things you'll need to know when choosing and using a software instrument. I'll also point out some important features and certain potential trouble spots that you should be aware of as you assemble your system.

TO PLUG OR NOT TO PLUG

Software instruments come in two types: standalone and plug-in. Standalone instruments are excellent — perhaps even preferable — for live performance. That's because they take complete control of the interaction with the drivers for your MIDI and audio interfaces, and this often results in lower latency. Plug-in instruments are often a better choice for desktop composition because they don't require intermediary software (such as Propellerhead's ReWire) for communication with your digital audio sequencer. It's not uncommon for a software instrument to come in both types, and in some cases the plug-in simply provides input and output to the standalone instrument.

Some plug-ins are native, which means they rely on the computer's processor for their operation; others require separate DSP hardware. Examples of the latter include Digidesign's TDM systems, TC Electronic's PowerCore, Creamware's Pulsar, or Mackie's Universal Audio Device (UAD). I'll concentrate on native formats here, but many of the same considerations apply to proprietary systems.

Native plug-ins come in several formats, but from a practical standpoint, your choices are limited by which host software you use (see the table “Plug-In Instrument Support”). Perhaps the best-known plug-in format is Steinberg's Virtual Studio Technology instrument (VSTi), which is supported by a number of hosts. On the PC, the other main choice is Microsoft's DirectX instrument (DXi), which is used by Cakewalk's Sonar among others. Some PC hosts support both formats. On the Mac side, in addition to VSTi, you'll encounter Mark of the Unicorn's MOTU Audio System (MAS), Digidesign's Real Time Audio Suite (RTAS), and for OS X users, Apple's Audio Units (AU). Because MOTU now supports Audio Units, MAS will probably fade away as OS X takes over the Mac world.

RACK 'EM UP

The first task in using a software-instrument plug-in is to get it installed and accessible to the desired host. That business is usually handled automatically by an installer that is supplied with the plug-in. It's important, however, to know where the installer puts the plug-in. VSTi and DXi plug-ins are typically placed in a folder in the same directory as the host application, and you can often sidestep a second installation for a different host by copying the plug-in to a new location. MOTU Digital Performer and Digidesign Pro Tools plug-ins reside in folders inside the System's Extensions folder. In Mac OS X, all plug-ins live in shared library folders where they are accessible to all of the appropriate hosts.

FIG. 1: In Steinberg's Cubase SX, software instruments are inserted into a virtual VST instrument rack (top right). The instrument's control panel can be opened for editing (center), and the instrument's output appears at a channel strip in Cubase's Mixer (lower left). MIDI for playing and automating is directed to the instrument through a MIDI track (top left).

Software instruments are handled differently by different hosts. Some hosts (Steinberg Cubase SX and Cakewalk Sonar XL, for example) have an instrument rack into which you insert instrument plug-ins (see Fig. 1). Once a plug-in is in the rack, it becomes available to sequencer tracks as a MIDI destination and an audio source, meaning that MIDI tracks can send their data to the instrument, and the instrument's output can be directed to an audio channel.

Other hosts (Emagic Logic and MOTU Digital Performer, for example) require you to add the plug-in as an insert in an audio track, or more precisely, as an insert in the mixer channel strip for that track. In Logic, MIDI and automation data reside on the audio track, and the audio output of the instrument is managed by the track's channel strip. In Digital Performer, a separate MIDI track is used to route MIDI and automation to the instrument, and the instrument's audio output appears at the channel strip for the audio track.

FIG. 2: The outputs of Propellerhead ReWire slave devices appear as inputs to the ReWire master device. In Cubase, activated inputs (green LEDs) appear as mixer channels (left).

ReWire, though not really a plug-in format, gives you another way to use software instruments with a host application. In that context, the host becomes the ReWire master (which is always launched first), and the standalone software instrument becomes the ReWire slave. Outputs from the software instrument appear as audio inputs to the host, and when supported, MIDI can be transferred in both directions (see Fig. 2).

LATENCY

Minimizing the latency of a software instrument (which determines its responsiveness to input from an external MIDI controller) is often a complex process. The greatest influence on latency is the audio buffer size. That setting determines how much audio is stored by the audio-interface driver before the audio is sent to the output. Large buffer sizes reduce the processor load; small buffer sizes reduce the latency. How to set the buffer sizes depends on your audio interface and the quality of its software drivers.

The computer's built-in audio drivers for all but the fastest laptop systems (and for many desktop systems, as well) seldom produce acceptable results. You should therefore plan to use a separate audio interface. Your MIDI interface and drivers may also affect latency, especially when intermediary software such as OMS or FreeMIDI is needed for routing MIDI from the host to the plug-in. That's the case with MAS and RTAS plug-ins as well as when linking standalone software instruments to a host.

AUTOMATION

Automation is an important consideration when working with software instrument plug-ins. It lets you save and reproduce various parameter changes (such as volume and pan) that occur during playback and mixing. For software instruments, automation capability may also include any or all of the instrument's front-panel controls. If and how each of the controls is implemented depends on both the host and the plug-in.

Automation comes in two forms: real-time automation, which typically involves some form of MIDI control, and after-the-fact automation, which is usually accomplished by graphically editing previously recorded MIDI automation (or by entering new automation data).

MIDI control allows sliders, knobs, and buttons on hardware MIDI devices (such as control surfaces and keyboards with assignable controls) to change a software instrument's settings. All hosts pass standard MIDI control data such as Mod Wheel, Pitch Bend, and sustain pedal to software instruments; but not all hosts pass along all the other MIDI controllers.

That might be because the host may use some MIDI controllers for other purposes — MIDI volume and pan for control of the mixer channel strip, for example. Another reason is that the host may reserve blocks of controllers for other plug-in slots on the same channel — for instance, to control other effects plug-ins.

Even if all MIDI data is passed to the software instrument, the software instrument may not provide access to all of its controls. The host manual's automation section will generally tell you which MIDI data is passed to software instrument plug-ins, and the software instrument's manual will usually have a table of MIDI controller assignments. Many software instruments have a MIDI-learn function for freely assigning any MIDI controller to any software instrument parameter, and some instruments even sense the type of MIDI controller being used — endless rotaries or 14-bit controllers, for example.

If you have an appropriate MIDI controller and can set up MIDI control of a software instrument, you can program and edit at least some settings using MIDI. Beyond that, automation requires that you be able to record those changes. That amounts to recording incoming MIDI data, and it's done in exactly the same way as when you record MIDI notes to be played by the software instrument.

Plug-In Instrument Support

Host	VSTi	DXi	AU	MAS	RTAS	Proprietary	ReWire
BIAS Deck	•					•
Cakewalk Sonar		•					•
Digidesign Pro Tools Free					•
Emagic Logic			•			•	•
Image-Line Fruityloops	•	•				•	•
MOTU Digital Performer			•	•			•
Steinberg Cubase SX/SL	•					•	•
Synapse Audio Orion	•	•				•	•

The situation becomes more complex, however, when you must modify existing automation. Unlike punch-in replacement for notes, in which all notes between the punch points are replaced, with automation the host has to sense which control data to replace. Some hosts offer sophisticated options for doing that; others expect you to manually edit out the automation you don't want. Some hosts don't provide for that at all and delete all existing automation when a new recording is initiated.

Another way to record automation is by changing the software instrument's onscreen controls. The software instrument determines what data is sent to the host; some send everything, others send nothing. The host determines how and where the data is recorded, but all hosts provide for that in some way. As with MIDI, replacing existing automation is handled in various ways.

The third method for entering and editing automation data is manually, using one of the host's MIDI editors. Often there is a graphic editor that displays the automation data as points connected by lines. You can typically add points, move them around, and sometimes even change the curve of the lines connecting them.

One thing to note about the point-and-line format is that it can generate a huge amount of MIDI data. The lines represent smooth transitions between the points, and entering a few points can result in a lot of automatically generated MIDI messages to create the transition. Multiply that by the number of controls being automated, and you can quickly clog the MIDI pipeline.

FIG. 3: Most software instrument hosts let you view and edit automation in several ways. Logic is typical in providing a list of MIDI messsages (top left), a graph (top right), and a bar chart with several editing tols (bottom).

Other editor formats that can be useful in editing automation include a time-stamped list in which individual points can be edited numerically and a bar-chart display that shows individual events and allows you to change their contour without adding data (see Fig. 3). Editors that are combined with a MIDI transformation processor (as in most sequencers) can be very useful when editing automation. For example, they could let you thin, quantize, scale, and offset automation data without having to deal with individual events.

RENDERING

Unless you plan to restrict your use of software instruments to live performance, you'll want to record each instrument's output, and you'll most likely want to do that within the same computer. That process, called rendering, is available in most software-instrument plug-in hosts and is often built in to some instruments.

The rendering process can be handled in a number of ways. Some hosts make you record all notes and automation before rendering to audio, whereas others let you render on the fly. Most hosts feature offline rendering, which produces an audio file from one or more tracks of your song much more quickly than it takes to play it. Rendering saves processor power because playing back an audio file demands significantly less of your processor than playing a software instrument.

Logic has introduced a new twist on the rendering front called freezing that renders an audio or software instrument track to a temporary audio file, then deactivates the original track and all its effects. Freezing tracks that you're not actively working on enormously reduces processor load, greatly stretching the track count of a laptop studio. Undoubtedly by the time you read this article, other hosts will be offering this feature, but if yours does not, you can accomplish the same thing by rendering then putting the rendered audio file on a new track and deactivating the software instrument (without deleting its MIDI data).

For standalone software instruments, the situation is slightly different. You either need to use an intermediary (such as ReWire) to pipe the software instrument's audio into your audio recording software, or you must use the software instrument's built-in recording features if it has them. The latter method is less convenient and requires sample-accurate MIDI synchronization, but it's often the only choice. And for recording on the fly, it might be preferable.

Although I've left the details for you to sort out for your particular software, you should now have a pretty good overview of what is possible with software instruments, how to get the most out of them, and what problems you might encounter. The number and variety of software instruments is enormous, and many high-quality products are free while others are very inexpensive. They may not yet replace all the hardware instruments in your studio, but they can certainly add to your collection, and for working on the road, they are indispensable.

---

Len Sasso can be contacted through his Web site at www.swiftkick.com.