Steinberg Cubase VST

This year's show‑stealer at Frankfurt's Musik Messe, Steinberg's Cubase VST isn't just another 'MIDI plus Audio' sequencer — it adds the effects and processing options hitherto only available in expensive TDM systems, for the same price as the previous, non‑audio version of Cubase. Is this the perfect desktop recording environment? Derek Johnson & Debbie Poyser find out.

Software sequencers are in a state of transition. When audio was introduced to MIDI sequencers, it was seen as a premium addition for top‑of‑the range packages. Recently, however, audio has started to establish itself as standard even on lower‑cost software. Opcode's Vision, formerly a MIDI‑only sequencer, now features audio, though in a more basic form than the company's own flagship, Studio Vision Pro. And with the launch of Steinberg's VST (Virtual Studio Technology) for Power Macs, audio now figures even in the standard Power Mac version of Cubase and Cubase Score, as well as in Cubase Audio. Anyone buying Cubase for the Power Mac now buys VST, an audio and MIDI sequencer which exploits this machine's 16‑bit audio capabilities, though for those who don't want audio or don't have a Power Mac, there's a non‑VST disk set too.

So does VST simply bring audio recording to the parts of the Cubase range which didn't previously have it? As welcome as that would be, given that the price has remained the same, VST actually adds more. In short, it takes the sophisticated MIDI sequencing of Cubase, plus most of the audio recording and editing facilities of Cubase Audio XT (excepting Digidesign TDM support), and adds comprehensive parametric EQ and four virtual effects processors — reverb, chorus, autopanner, and stereo delay. You can now sequence MIDI parts to your heart's content, manipulate them with powerful editing facilities, then add a number of channels of audio material, which can be EQ'd, effected, and mixed — all within your Power Mac, with no need for an audio recording card, or external effects or EQ. The final track is then output from the Mac's stereo output for mastering. The only external gear you'll need is MIDI instruments, a mixer (more later on this), an amp and speakers, and/or headphones.

Surely there's a catch? Well, Power Macs' audio connections are on stereo mini‑jack sockets, not the most robust or high‑quality of audio connectors. Its stereo input means that you'll only ever be able to record two audio channels at once, and the stereo output means that your finished work will have to be mixed inside the computer. You also have to rely on the Power Mac AD/DA converters, which won't be audiophile quality. In a pro studio, these would be real limitations, and you'd want to spend the money on additional hardware to ensure the highest audio quality.

However, not all of us have pro studios. Thousands of musicians have modest 'project' setups, where they produce demos and even budget releases. These are the kind of people who'll be interested in Virtual Studio Technology.

Hardware & Installation

The first requirement is a Power Mac, running System 7.5 or higher (but not 7.5.3 at the present time). Steinberg recommend the 90MHz 7200, and the minimum machine they say you should use is the 6100. The speed of your computer is one of the factors that governs how many audio channels you'll be able to play back, and how many simultaneous EQs and effects you can use. Other factors are hard disk speed and installed RAM. The minimum free RAM for VST to run is 7Mb (it won't even stoop to installing with less than 16Mb total RAM!), but the more RAM, the more audio channels you can have. You need a 256Kb 2nd level cache memory, which should cost around £120‑150, and some kind of data backup method will be required.

Installation is easy, using Master Disk Authorisation copy protection, and the next step is deciding how many audio channels to assign to your system. This figure is quite fluid, depending on your hardware and memory, but Cubase helps you set a realistic number. Enter a number of channels in the Audio System Setup (pictured above right) dialogue and set an amount of memory per channel from a preset list — the more the better, as far as audio performance is concerned. You also have to set a disk block buffer size, which governs the buffer size used when reading and writing data from/to the hard disk; the two figures are intimately related. If you set too many channels for the RAM available, VST alerts you. Then you have to juggle channels and memory allocations.

According to the manual, a Power Mac 7200/90MHz machine with enough free RAM for the recommended memory assignment (around 12Mb), working with its internal drive, should be able to handle 16 audio channels and 32 activated EQs. With a 9500/133 and a large, fast hard drive, you might get as many as 32 channels and 64 EQs. On our 75MHz 7200, with 16Mb of RAM, we were able to run six channels of audio with two EQs each and all four effects processors. We didn't even assign the recommended amount of RAM, giving VST 10Mb instead of 12Mb.

An 'Audio Performance' window (shown left) with CPU and disk load meters can be left onscreen at all times, and if you're overdoing it for your system, red Over indicators light up. However, by then you may already have heard the distortion which means you need to pull something back — turn off an EQ, for example.

Getting Started

First, plug a sound source into the Mac's input. Many musicians already have a multitrack mixer, and will feed sound sources into the Mac via that — which is advisable, because the Mac's input is set up for mic‑level signals and is quite fussy about incoming levels. You'll probably have to make a special lead to connect your mixer to your Mac; we noticed that our Power Mac's socket recesses are slightly too small for the average mini‑jack to fit perfectly. The plug appeared to fit, but produced unusual levels of noise because of incomplete electrical contact. We were dismayed at this noise until we diagnosed the problem. You may have to obtain mini‑jacks with extra‑skinny barrels, or carefully shave a little off the ones you have. Even with a correctly‑fitting connector, there is some hiss from the Mac output, but at normal volume, it isn't even as bad as the average synth or effects processor.

Recording is initiated from the Arrange window. The Track Class menu allows you to select an Audio track; opening the Track Inspector box then reveals Monitor and Record Enable buttons. Record Enable readies a track for recording, and you then select an audio channel to record with. The Mac assigns signals from the left input to odd‑numbered channels and those from the right input to even‑numbered channels. Stereo or other two‑channel tracks can be recorded: set their channels to 'Any' in the Inspector.

Record Enabling also enables monitoring for the selected channel. VST offers three monitoring types: Tape Type and Record Enable use the Mac's hardware for monitoring, but its design creates a buffering delay between the input signal and the same signal being monitored by VST's mixer. If you're recording 'live' audio, like someone singing, this can be enough to throw their timing, and in this case, it would be better to select the 'Global Disable' option. This turns off Mac monitoring and lets you use your own mixer for monitoring instead.

To help you set the correct input level, VST has a nice‑looking Monitor Mixer (shown above left) with a Master Fader window (shown left) whose meters can show signal input level. This should not exceed 0dB: avoid going into the red and lighting the clipping LEDs. If levels are too high, your instinctive reaction will be to pull down the on‑screen faders. However, this won't help, since you cannot set input levels within VST: you have to do it at source.

The Monitor Mixer is also used for final mixdown. It can be automated via a separate 'mix' track, which records all mixer movements, including mutes, solos, and EQ and effects changes. Complicated mixes can be assembled a track at a time, and you can even edit individual mix events using List Edit.

Recording Audio

When you enable recording for a new Song, VST asks where to store the audio Files. You can't create a new folder at this point, so make a habit of creating one for each project before starting to record.

With input level set, recording can commence. A selectable Pre‑record parameter starts recording early to catch the attack of sounds starting before a bar‑line, and if you fluff part of an otherwise good recording, Punch‑in (programmable or manual) is easy to set up and produces flawless results. During recording, VST acts very much like tape, and the usual transport controls navigate around an audio track, but it also has some great extra features. Cycle Audio, for example, allows you to set up a loop of part of your song and record multiple consecutive takes of the same part. In the Audio Editor, the takes are ready‑stacked, and you can play them back individually to find the best parts of each, for assembling a composite take. This is great when recording vocals or guitar solos. When you've identified the best bits, you can delete the rest, leaving portions of audio scattered across various takes. Selecting 'View by Output' then shows them all together, looking like a single take.

VST lets you record as many audio tracks as you want (disk space permitting), one or two at a time. What you can't do is play back as many as you like at the same time. The important restriction is audio channels. Tracks are played back by channels, and if you only have eight audio channels in your system, that's all you'll be able to play back at one time. You could play back more, however, by using the channel space cleverly. For example, if one track on channel 2 contains a vocal which occurs during choruses, use the space in between for another track containing a guitar part which occurs during verses. The guitar track would also be set to channel 2, and the result would be two tracks playing back on one channel. If the audio on those tracks overlapped, however, only one track would sound, as they would be fighting for the same channel.

Another way of stretching available channels is Audio Mixdown — like 'bouncing down' with multitrack tape. It re‑records the contents of multiple tracks playing on several audio channels onto another track playing on one channel, freeing up the original channels. With VST, you don't even have to keep a spare channel to bounce onto, because as soon as the original tracks are muted or deleted, their channels become available for playing back other tracks. Audio Mixdown bounces can even be in stereo, with all level, pan, EQ and effects settings intact.

Audio recordings show up on the Arrange page alongside MIDI tracks. You can use many of the same tools you'd use on MIDI data to move, copy, split and paste the audio, but for more detailed editing, you'll probably use the Audio Editor.

The Audio Editor

This is the best place for audio track trimming and assembly — cutting a long recording into smaller Segments, masking parts of the audio with 'Inset Handles', editing volume and pan curves, and working with 'Q' (Quantise) and 'M' (Match) points.

Cubase has powerful features for quantising audio events in a similar way to MIDI events. However, since audio, unlike MIDI data, doesn't automatically contain reference points, the audio quantising procedure is inevitably more complex than MIDI quantising, since reference points have to be created. Cubase inserts 'Q‑points' at sensible locations when you record, and when you perform operations such as splitting audio into smaller parts and manually punching in. There is only one
Q‑point per audio event, and they are used by the program to snap events to musical positions when you move audio, and also for quantising. Though Cubase inserts them, you can edit them. Match points are markers inside an audio event, which you or Cubase can generate. There can be many of these for one event; they should correspond to significant points, like beats in a drum part, and are used for creating audio groove templates, and Match quantising of audio to tempo, audio parts to MIDI parts, or even audio parts to other audio parts. These audio quantising features would be invaluable in certain situations, and it's nice to know they're there, but they are complex to use.

One operation you're likely to perform is deleting silence from an audio part, to save disk space. There is a special function for this, called Banish Silence, which is not very well explained in the manual. The waveform appears in a dialogue box with a pair of horizontal and a pair of vertical lines superimposed on it. You can move these lines so that they set a time and level threshold, selecting a 'slice' of the waveform that you consider corresponds to silence. Cubase then finds sections of waveform fitting the description and nukes 'em. However, the waveform display in the dialogue is small and a bit fiddly to use. We thought it more fun to simply blow the waveform up in the Audio Editor, go through it finding silent bits, which are quite obvious, and then deleting them. This doesn't take long for an average track. Deleting is done using the Backspace/Delete key, which removes the selected area of waveform from the display, or doing the same while holding down Command, which deletes it from disk too. Audition the waveform here by clicking on it with the magnifying glass, and scrub audio by dragging the mouse over the waveform in either direction. Auditioning is great, but scrubbing sounds like Pinky and Perky on speed! To edit several audio parts at the same time, selecting them in the Arrange page and then double‑clicking on one of them displays all their waveforms in 'lanes' in the Audio Editor.

The Wave Editor

Audio Editor operations are non‑destructive of Files on disk but Wave Editor operations work directly on the audio File. You can Undo your last operation, but going further than that will change the audio File permanently. So if you want to experiment in the Wave Editor, duplicate the File first and work with the duplicate. Once in the Wave Editor, you can zoom to single sample level, for detailed editing, and apply DSP functions: these include Fade‑in and Fade‑out of events; creating Crossfades between events; Inverting Phase; Normalising (to boost a quiet recording's level); Quietening (to reduce the level of a recording); Reversing; and Silencing (replaces a portion of audio with silence).

Rather more fun Wave Editor features are Pitch Shift and Time Stretch (see screen window on page 94). Briefly, Pitch Shift changes the pitch of an audio File without altering its length, and Time Stretch alters length without affecting pitch. Pitch Shift can change the key of a whole vocal, say, or just fix wrong notes, and it produces surprisingly good results as long as you make small changes. A shift of one semitone up sounds natural, and two semitones would probably be OK on a backing vocal in a mix. A downwards shift of two semitones is quite usable. Using Time Stretch for moderate tempo changes is effective too, but note that neither of these processes works terribly well on long stretches of complex audio.

More sophisticated pitch‑shifting and timestretching is available from Steinberg's Time Bandit (an early version of which was reviewed in SOS April '94). If you have it installed, it's easily accessible from within VST; a couple of VST's audio and tempo matching operations actually require Time Bandit.

The Pool

When you've made several recordings, the Pool, a kind of database for keeping track of recordings, comes into its own. On opening it, you'll see a list of audio Files. Clicking on the arrow next to a File shows its 'Segment' list. Every File has at least one Segment, which is an instruction to play a portion of the File. A Segment could play a whole File, or just part of it, and when you chop a recording into smaller bits, they become Segments in that File's Segment list. Because various operations create new Segments, the Pool can become crowded with Segments which may not be used in the current Song. The 'Purge Segments' function finds Segments which are not referenced, and deletes them from the Pool — although this doesn't affect the audio Files on disk. The Pool offers other File and Segment deletion functions, as well as showing where Files are stored, when they were recorded, their length, start and end points, mono or stereo status, and how often they're used in the Song. File renaming should be done here: if you rename Files in the Finder, Cubase may not be able to locate them next time the Song is opened. Though you can manually re‑locate Files and tell Cubase their new names, it's best to avoid the problem altogether. It's also possible to drag Segments from the Pool into the other editors.

Administering recordings is easier if you understand what happens to Files when you delete them. To tidy a messy Pool, without actually deleting audio Files off disk, use the Delete key. Using Command‑Delete removes Files from the Pool list and the hard disk — only do this if you're sure you don't want them. Another command to handle with care is Delete Unused Files: this finds Files not used in the current Song and removes them from the Pool and the disk.

The relationship between Files and Segments isn't too hard to comprehend (Files are full audio recordings, saved on disk; Segments are simply specifications for playing sections of Files), but there's another term to get used to: Audio Events. You could compare these to MIDI Note On messages for audio, since they tell a Segment where to start playing. One operation that illustrates the distinction between Audio Events and Segments is making an Audio Event play back a different Segment to the one it's currently playing. Clicking on an Event in the Audio Editor, while holding down Command and Option, brings up a menu listing all the Segments in that audio File. You can then choose another Segment for the Event to play — a verse vocal Segment could be played back by the Event that's currently playing the chorus vocal Segment. Fortunately, Audio Events are pretty much invisible to the user, and you can usually ignore them and concentrate on working with Segments.

Effects & EQ

Apart from support for Power Mac audio, Cubase VST's other brand‑new features are its four effects processors and comprehensive parametric EQ (shown left). Steinberg are encouraging third‑party manufacturers to develop VST plug‑ins, so more effects and processors should be available soon.

VST's effects are accessed from the Monitor Mixer, where four effects 'sends' route pre‑ or post‑fader signal from each channel to the effects. They're presented in a 'rack', and are very nicely styled, with metallic finish and 3D controls. The effect types are:

• Espacial, an editable reverb.
• Choirus, a chorus/flange hybrid, also fully editable.
• Stereo Echo, a simple programmable stereo delay.
• Auto‑Pan, also programmable (see box for effect editing parameters).

Effects can be configured as four of one type, two of two types, or any combination, to a maximum of four processors. Equally welcome are the banks of preset effects, and the ability to save custom effects, individually or in banks.

Speaking subjectively, the quality of the effects is easily as good as standard hardware processors, and all changes can be heard in real time. It's tremendously useful and very intuitive to have effects on tap within an audio sequencer, and now those of us who can't afford a TDM system can feel the benefit! Having four processors gives plenty of scope, but remember that the Audio Mixdown feature mentioned earlier records all EQ and effects settings for audio tracks, thus freeing up effects for 'live' processing — like recording effects to tape in a traditional studio.

Each band of EQ has a frequency range of 20Hz‑16kHz. Steinberg have thoughtfully provided four presets — Lo, Lo mid, Hi mid and Hi — for instant access to four bands of EQ in useful frequency ranges. Control is also provided for gain (+/‑12dB), centre frequency, and bandwidth. Once again, presets are provided for bandwidth, from wide to very narrow. The EQ is very good, and is capable of being as drastic or as subtle as you want, with all results in real time.

The Good & The Bad

We found odd problems when working with VST, but none were insurmountable. Most difficult, perhaps, was getting used to how careful you have to be with Power Mac input levels to avoid distortion. You also have to be a little careful with playback levels. You can boost a signal by up to 6dB, and this is generally plenty, but you may get the odd crackle of distortion if you boost all your audio channels at the same time. Just pull back a fader or two and it cures the problem.

Other niggles include the fact that EQ setup pages are usually accessed from the Monitor Mixer via 'FX' buttons — most people would intuitively click on the adjacent 'EQ' button, but this only works the first time you select the EQ page for that channel; subsequent clicking on the EQ button simply turns EQ on and off. To add to the confusion, the manual states that the EQ button opens the EQ page, while an accompanying diagram indicates the FX button! Also, setting exact values in the effects window with the mouse is incredibly difficult; the manual gives a technique for fine control over parameter dials, but this simply doesn't work well enough on the effects display, and you can't simply type a value into the display window.

In places, it feels as though the loose ends of the audio side aren't completely tied up. For example, the manual says audio tracks can't be put into Group Tracks (as MIDI tracks can) but if you try to anyway, the software lets you go almost all the way there but won't complete the procedure. There's not even an alert to say "Not with audio please!" And even if you save a Song before quitting, when you close down the software, a dialogue asks if you want to save your Song — Cubase doesn't sense that you've just done it. Another irritating point is that the manual says explicitly that Cubase automatically groups stereo recordings: it doesn't, which can be vexing when you move them about and only one half of the stereo track responds, thus throwing out your track positioning. Sometimes, the Wave Editor temporarily refused to audition processed waveforms, though it soon recovered on each occasion, and finally, using Mixdown (not Audio Mixdown — see the 'Manuals' box for the distinction) to group several tracks onto one track was unpredictable in operation. It didn't always mix down all the chosen tracks, affecting perhaps three or four out of five selected, for example. This point has been passed on to Steinberg, who are now working to fix the anomaly.

There are many nice touches which balance the above, though: audio continues playback during other operations, even when you're changing effects and EQ settings; faders in the Monitor Mixer can be linked in adjacent odd‑even pairs by holding down the Option key while moving one fader, and the Option‑linking of channels usefully extends to effects and EQ settings; Power Mac native code means that VST runs very fast even on our 75MHz machine; and the sound quality of VST audio recordings (16‑bit, 44.1kHz sample rate) is subjectively rather good. VST also seems stable: during our time with the software, it only crashed once.

Conclusion

The bottom line is that Steinberg have added powerful Cubase Audio‑like features, plus an as yet unprecedented selection of digital effects and EQ, to their basic MIDI sequencer for a starting price of £329 — and that's simply amazing. For integrated MIDI sequencing and audio recording on a Power Mac, there is nothing as cheap with a similar feature set. Although the Power Mac's audio hardware is not as good as you would get from a Digidesign board, the audio quality obtainable with VST is easily good enough for demos, and exceeds the performance of cassette‑based multitrackers.

Many a home studio‑based musician will have been looking forward to such a cost‑effective, all‑in‑one solution to MIDI sequencing and digital audio, and in this context, VST is a great success.