The Freedom Factor
Opcode's ground-breaking sequencer, Studio Vision, lets you manipulate audio data as easily as MIDI events. All you need is a hard disk recording system like Digidesign's Sound Tools, a fast Macintosh computer with plenty of storage capacity, and some creative ideas. Paul D. Lehrman reports.
When MIDI sequencers first came on the scene, composers were delighted with the freedom they granted. The ability to correct mistakes quickly, to edit one musical parameter in a phrase without affecting all the others, to quantise rhythms, to move whole sections around in an instant, to preserve and choose from an infinite number of alternate takes — all the features sequencer-dependent musicians now take for granted were a godsend for those of us accustomed to working with multitrack tape.
The one thing often missing from a MIDI studio, however, was 'live' sound. For various reasons, many sounds are beyond the capability of MIDI instruments to produce convincingly: human speech or singing, screaming sax solos, romantic violin cadenzas, and the intricate performance techniques of many non-Western instruments. For composers to use those sounds meant putting them on tape. Syncing to tape has always been an essential feature of most MIDI sequencers, but the fact that part of the piece was now committed to a linear ribbon of oxide instantly re-imposed many of those limits that sequencers had freed us from. Once something is on tape, shifting it around in time is still difficult, and because you can't easily splice timecoded tape, it's even more difficult than in pre-MIDI days.
One partial solution has been to use samplers for 'live' sounds, and trigger them from a sequencer. This works well for short sounds and phrases, but dealing with longer forms requires very expensive hardware, as well as tons of patience, since samples are normally recorded, edited, transferred, and stored in separate, time-consuming, operations.
In the last two years, we've seen the emergence of low-cost, personal computer-based hard disk recording systems, in which audio signals are digitised and stored in real time on a Winchester disk, under the control of an off-the-shelf computer like an Apple Macintosh, Atari, or IBM PC. Hard disk audio offers many significant advantages over tape, including linear frequency response, minimal distortion, freedom from wow and flutter, ability to copy or bounce a nearly infinite number of generations without building up noise, random access, and the ability to mix and process entirely in the digital domain without the attendant problems of constant analogue-to-digital-to-analogue conversion.
The computer muscle required to handle MIDI, at its 31,250 bit-per-second bandwidth, is almost trivial compared with what's needed for hard disk audio, which has to deal with (44.1 kHz x 16 bits x 2 channels =) over 1.4 million bits per second. Therefore, a couple of hard disk recording systems have appeared that include MIDI capabilities. Digidesign, makers of the Sound Tools and Audiomedia systems for the Mac and Atari, have been at the forefront of this movement. Their Deck system [reviewed Sound On Sound, November 1990] and Q-Sheet A/V software allow both hard disk audio and Standard MIDI File playback. In addition, Deck allows audio recording and a modicum of MIDI recording and editing.
But the first product to combine MIDI and hard disk audio in one comprehensive creative tool is Studio Vision, from Opcode Systems. Studio Vision takes the opposite approach from Deck: it enables Vision, Opcode's topflight MIDI sequencer, to also handle digital audio recording, editing, and playback.
Studio Vision looks just like Vision when you boot it up, except it has an extra menu, 'Audio'. It tries to make audio manipulation as easy as MIDI manipulation, and to bring many of the advantages of MIDI-based composition to audio editing. It's a simple, logical idea — but it's also revolutionary, and best of all, it works!
Studio Vision, which was created with lots of technical assistance from Digidesign (who are located literally next door to Opcode in California's Silicon Valley), is a very hardware-intensive application. It requires a Macintosh II or SE/30 computer, plus Digidesign's Sound Tools or Audiomedia. It will run, theoretically, with two megabytes of RAM in the computer, but if you want to use MultiFinder you need four megabytes; if you want to take advantage of certain audio editing features, make that five. It is MIDI Manager compatible but, as I'll explain later, to run it under MIDI Manager you'll need a very fast computer. You also need a hard disk — as with all disk recording systems, the bigger the better.
Audio files, or 'soundfiles', can be imported into Studio Vision from Sound Designer, Audiomedia, Dyaxis's MacMix, Passport's Alchemy, or any other AIFF (Audio Interchange File Format)-compatible application. (It will not directly accept 8-bit Mac Recorder-style files, but various conversion programs are available.) When you import a soundfile, you assign it to a track, tell it where to start playing, either in musical (bars/beats/'units') or SMPTE time, and assign it a master volume, from 0 to 127. In keeping with the MIDI orientation of the program, this is called the file's 'velocity', although of course there's no real keystroke involved.
Studio Vision also has its own recording capabilities, and you can record in mono or stereo. You choose a track, open up a Record Monitor window which contains two bargraph level meters, give the recording a name (stereo files get two names), tell it what disk to put it on, and click on Record. While you are recording audio, you can play back any MIDI tracks already in the sequence, the program's metronome, and/or a previous audio track. You can record using the Mac's internal clock, or synced to MIDI Time Code.
When you're done, click Stop. The program takes a few seconds to draw a picture of the recorded audio waveform, and then displays it in an audio 'track' at the bottom of the sequencer track's graphic window. If you want to do another take, click Record again. If you want to save disk space and throw out the first pass, select Undo Record and start again. You can disable the waveform drawing if you like, which can save you some time, but you'll need it later for most editing functions.
Studio Vision's track structure is flexible enough so that not only can multiple MIDI channels be combined on one track, but also MIDI and audio — in fact, multiple audio tracks — can co-exist on a single sequencer track, and be viewed in one window. The program provides up to 16 'audio instruments', any of which can be assigned to as many sequencer tracks as you want, and a soundfile, mono or stereo, can be assigned to any instrument. If you have a colour monitor, you can assign a different colour to each instrument — talk about an orchestral palette!
These audio instruments (unlike Vision's MIDI instruments, which are hardware and MIDI-channel specific) are actually just software designations. Any instrument can play at any time, and can be panned anywhere in the stereo field. Once an audio track is in a sequencer track, it can be moved around in time just like a MIDI event. If you start playing a track in the middle of an audio event, the audio will pick up right at the proper place, as if it was just another part of the sequence.
The sole limitation is that only two audio instruments (only one if it's stereo) can be sounding at a time. If that limit is exceeded, the last soundfile to be triggered will take precedence. This is not as much of a problem as you might think, because the program includes extensive audio mixing functions which do not destroy the original tracks, and so remixes can be redone at any time, assuming you have enough hard disk space to store all the various versions intact. These functions are all performed off-line — that is, first you listen to what you want to mix, select the tracks and regions you want to combine, and then tell the computer to do its thing. The program creates a new soundfile in a few moments.
You can impose 'controllers' on an audio track, much as you can on a MIDI track, by recording them from a MIDI source, or by drawing them in on the track's 'Strip Chart', or using the program's on-screen faders. You can enter any controllers you want, but the only ones that will have any effect are Volume (controller number 7) and Pan (10). As with a MIDI track, you can hear the controllers' effect on the sound as you record them. Each audio instrument on a track can have its own set of controllers. When you mix down audio tracks, you can have these controllers affect the mix, so a mixed audio track can contain as many 'moves' as you like.
Editing audio tracks is highly logical and simple with Studio Vision. Using various cursors, you can cut, copy, paste, and drag around sections of tracks with ease, and hear what you're doing immediately. You can change which files are assigned to which instruments, so that different combinations of files sound at a particular time (two files sharing the same instrument cannot sound simultaneously), and you can move audio tracks to different sequencer tracks, or different sequences. You can clone a file and make it appear somewhere else (or even just echo itself) by Option-clicking on it.
SMPTE numbers are always available on the screen for reference, and you can go freely back and forth between graphic and edit list windows. All of this happens quickly because the program is not actually moving audio data, it is simply repositioning pointers. It's a wonderful way to edit audio in time, far faster and more intuitive than any other method I've ever encountered.
"Once an audio track is in a sequencer track, it can be moved around in time just like a MIDI event. If you start playing a track the middle of an audio event, the audio will pick up right at the proper place, as if it was just another part of the sequence."
For more serious editing and processing, Studio Vision provides a 'hot link' to Digidesign's Sound Designer II sound editing program. When you select an audio track or region and choose 'Edit Soundfile' from the Audio menu, Sound Designer II automatically opens up (under MultiFinder) with that file showing in the window. You can now perform any operation the program allows on the file — including pitch and time changing, merging, equalisation, compression, etc — and when you switch back to Studio Vision, the changes are in place. There is no copying through the Clipboard, or saving to disk: since the file is already on the disk, you don't need any of that — you simply have one file which two different programs can work on while it's open. (This, by the way, is going to be a feature of many programs which run under Apple's long-awaited System 7 for the Macintosh.) One very nice touch is that you can tell Studio Vision to close all of its many windows when you are using Sound Designer (or any other application under MultiFinder), which can reduce screen clutter considerably.
Be warned, however, that to take advantage of this feature you will need the latest version (2.0) of Sound Designer II, which means you will need a full Sound Tools system. The Audiomedia software doesn't support this function, although Digidesign say that an update that will is scheduled for release "imminently". Your Mac will also need at least 5Mb of RAM — 2Mb for Studio Vision, 2Mb for Sound Designer, and 1Mb for the System.
A soundfile can be divided into multiple sections for individual manipulation, using the Separate function, which can be very useful in a number of ways. For one thing, you can use multiple audio instruments for different takes of a recording, separate each track into individual sections (phrases or notes, for example), and then assemble a single perfect track by choosing the best sections from each take. Before you commit yourself, you can audition your choices using the 'Play Selected' option, and when you've made a final decision, merely copy the selected sections in one operation and paste them together into a single track. Again, because you're only manipulating pointers, the operation is quick and uses up no additional disk space.
A corollary to this is Studio Vision's most awe-inspiring feature: 'Strip Silence'. This is a kind of digital 'gate' in which you define what silence is — a certain level threshold, and a certain amount of time, in milliseconds, that the signal must be below that threshold — and the program then deletes those sections from the soundfile, and separates all of the remaining parts of the track into discrete events. Although the threshold setting uses arbitrary units, the screen provides clever visual feedback as you adjust it, which helps a lot.
The Strip Silence function is also Undo-able, which helps even more. This means that it is absurdly easy to edit the placement of individual words in a dialogue track, make musical phrases happen earlier or later without re-timing the whole track, pick out notes or small sections and repeat them either immediately or later in a song, or — and this is the best part — quantise discrete audio events. Sound effects can be lined up automatically with music, background vocals will never be late, that live bass track can groove perfectly with the sequenced rhythm track, and if you're working with the world's worst drummer, you can record him, separate the individual drum beats, and quantise the hell out of him!
All of this audio editing is totally non-destructive, and everything can be undone, either right away or days later. If the silence threshold you choose proves to be too high, for example, you can always go back to the original audio event and try again. Or you can take one of the separated events and change its end time — stretch it — so that it once again incorporates the rest of the file (you can't do this in the graphic window, however, you have to use the edit list window — which works fine but isn't nearly as much fun).
Also, these features can help you save disk space, which — no matter how massive a hard disk you are using — you never have enough of. Once you've done all the editing and separating you want, and you don't need your original files any more, you can eliminate the leftover audio from the original (that is, audio which no pointer is pointing to) with the 'Compact' function. How much disk space you will recover will depend entirely on the source material and how you use it, but I found I could often eliminate nearly 70% of the space used by the original.
Another very useful function, when you are using audio from many different original sources, is 'Consolidate', which takes all of the portions of the various files you are using and splices them together in one file. This doesn't save you any disk space, but it makes storage and retrieval a whole lot easier and faster.
As impressive as Studio Vision is, it's not without its teething pains. The program locked up about a dozen times in the first couple of days. This is hardly unheard-of for a new program, but it means you have to be careful to save your work frequently. Fortunately, no files — especially hard disk audio files — were ever trashed in any of these crashes, so the damage was minimal.
More serious problems are found in the area of timecode handling. Mixing audio and MIDI on the same platform brings up some interesting issues. If a MIDI sequencer is synchronised to tape (using SMPTE/EBU timecode converted to MIDI Time Code) and the tape runs off-speed, the sequencer will slow down or speed up. If the speed variation isn't too gross, the results won't be audible — after all, the pitch won't change.
On the other hand, Digidesign's hard disk audio systems (except for the very latest version of Sound Designer II) sync to SMPTE/EBU by starting to play when they receive a specific timecode frame number, and then continue using the Mac's internal clock as a timing base. If such a file is playing at the same time as a MIDI sequence and the tape is off-speed, the MIDI and the audio will drift out of sync. In Deck, this is not hard to make happen.
Studio Vision attacks this problem on two fronts. A Tape Calibration function makes sure that the program understands exactly what speed your tape recorder is playing at. You open up a window, start your tape, and tell the program to measure the speed of the tape machine. It compares the incoming MIDI Time Code messages with what its internal clock thinks should be the speed, and displays how fast or slow the tape is running, to the nearest 1/1000th of a percent. It averages the speed variation over the period of time the window is open, and then puts that number into the sequence. (Once it's in there, there is apparently no way to remove it, except by recalibrating. The manual is very vague on all of this.) Now when you record audio while the sequence is synced to tape, the timing of the audio events ('real' time) should match exactly the timings displayed by the program (SMPTE time).
The other tape speed-related function is 'Lock Audio To Tape'. This is used when you are playing back audio, with the tape machine as the sync master. It measures the instantaneous speed of the incoming MIDI Time Code, and changes the speed of the audio playback accordingly. Unfortunately, this function sometimes creates more problems than it solves. Unlike MIDI sequences, minor periodic speed changes in audio playback sound terrible, because the ear is very sensitive to pitch changes. When I hooked up a free-running SMPTE generator (Mark Of The Unicorn's Video Time Piece), which theoretically should output perfectly steady timecode, to Opcode's Studio 3 SMPTE-to-MIDI Time Code convertor, I would occasionally hear a distinct 'wow' imposed on the sound. It was especially noticeable on acoustic guitar and piano tracks. I also noticed it when I played back a VHS video tape striped with SMPTE through the Studio 3.
The problem proved inconsistent, so it was a little tough to pin down exactly what was going wrong, but my guess would be that the software algorithm used in the function is too sensitive. A spokesman thought that there must be something wrong — or at least "not professional" — with the SMPTE equipment I was using, but many small studios using Studio Vision will have similar low-cost timecode and video equipment, so I can't accept that as a valid argument. I was also told that Sound Designer II (Version 2.0) uses a similar locking scheme, so I tried it with the same file and hardware setup, but I noticed no speed variation.
Studio Vision is very picky when it comes to hardware, and it experienced a certain amount of trouble with a Syquest drive. The 'Your hard disk is fragmented or too slow' message came up a lot, even if I had just optimised the disk and had 15 megabytes free. Often it would just stop recording or playing back in the middle of a file, with no explanation — and the next time through it would be fine. Digidesign have been quietly telling people not to perform heavy-duty audio editing on Syquest drives, and this appears to be sage advice. The best setup for Studio Vision would therefore seem to be a large fixed hard drive for on-line work, and Syquest removable cartridges (or, if you can afford them, read/write magneto-optical discs) for off-line storage. Optical discs that will be fast enough to play from, and therefore edit on, but not record on, are apparently going to be available in the near future.
"All audio editing is totally non-destructive, and everything can be undone, either right away or days later... you can always go back to the original audio event and try again."
And one other hardware issue: If you try to operate Studio Vision with a Sound Tools system with the AD IN box turned off (because you're not planning to record anything during that session), make sure you let the program know about it, by using the Hardware Setup window. Otherwise you'll get error messages, crashes, and all sorts of disasters. It would be nice, and avoid a certain amount of hair pulling, if the program could detect the problem ahead of time and warn you that your current setup is not what it expects it to be. (The same problem occurs, incidentally, in Sound Designer II, but that program is polite enough to just not work.)
Studio Vision is MIDI Manager compatible, but performance under MIDI Manager with a Mac IIex was unacceptably slow — screen redraws took forever — and also subject to crashes. Apparently you must have a IIci, or even better a IIfx, to use it. This means if you have a slower machine (and whoever thought a 16MHz processor would be considered slow?), and you want to use Studio Vision along with one of Digidesign's NuBus-slot music cards — the MacProteus or the new Sample Cell, you're going to have to invest in an accelerator or a motherboard upgrade. (The new Macintosh IIsi is fast enough, but it has only one NuBus slot, so the situation won't arise. Dumb move on Apple's part, if you ask me.)
One minor limitation I would love to see changed is that you cannot use Studio Vision's new Tap Tempo feature (where you hit a Mac or MIDI key at any tempo you choose, and the program records the time intervals into a tempo map) when audio is playing. Opcode says this is because the program is too busy to accept any external timing changes when it's spitting out audio, which seems reasonable, but wouldn't it be nice to record a vocal track in free time (without a metronome), and be able to construct a tempo map for the accompaniment afterwards? See what you can do in a future version, gentlemen.
Finally, one suggestion: the Strip Silence function, good as it is, would be even more useful if, instead of simply searching for times at which there is no signal present, it included some slope detection. This would mean that it could determine the difference between attacks and decays, and be more forgiving when it comes to cutting off decays — in other words, it should act more like a real gate.
It should also be mentioned that Opcode's Vision sequencer has undergone a few changes lately, and Studio Vision incorporates all of these changes. Both programs are now shipping in Version 1.21. Besides the aforementioned Tap Tempo function, the major new features include:
• The program can 'subscribe' to a program list 'published' by Opcode's universal patch librarian, Galaxy, so that instead of program numbers appearing in the program change windows, program names can be displayed.
• All 32 software faders can be displayed simultaneously, with their sources or destinations, and the faders follow playback.
• Graphic windows scroll when a sequence plays.
• You can hear notes sound when you click on them in a graphic or edit list window, and a scrubbing function is available for listening to a track's notes as you move the cursor.
• Several curves as well as freehand drawing are now available in the Strip Chart window, so controller movements are much more easily plotted.
• The notorious MIDI Time Code two-frame bug, in which the program was always two frames late, has been fixed.
The timecode issues are probably not going to affect many users, and I imagine they're easy enough to fix. The other gripes I'm sure will be dealt with in an upgrade or two. So when I'm done complaining, what do I think of Studio Vision?
In a word, it's brilliant. A stellar concept, executed extremely well. If any one product manages to kick MIDI-produced music out of the repetitive rut it often falls into, and gets project studios working with live musicians again, this will be it. (It should also convince a lot more British studios to start using MIDI Time Code.) Even without its MIDI sequencing capabilities, Studio Vision brings some revolutionary and very welcome ideas to audio editing. Besides its obvious musical applications, and although Opcode do not seem to be pushing it in this realm, I see it as a marvellous tool for sound effects, dialogue editing, and other non-musical and visual-orientated tasks.
Studio Vision is easy to use, especially if you already know Vision. If you don't, it's not the fastest sequencer to learn, but it's worth the effort. In all, a truly visionary product (ahem!), which provides a clear vision (stop that!) into the future of computer-based music making.
(Thanks to Dave Mash, of the Berklee College of Music, for his assistance in preparing this review.)
Paul D. Lehrman is a composer, author, and teacher who lives in Massachusetts, USA. He has recently upgraded his memory capacity, and reports that all his slots are now full. He would like to do the same with his computer next.
£795 inc VAT.
MCMXCIX, (Contact Details)
Review by Paul D. Lehrman
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!