I last reviewed Mark Of The Unicorn's (commonly abbreviated to MOTU) Performer MIDI sequencer way back in the November 1986 issue of Sound On Sound. The intervening 10 years involved a long relationship with Atari, but I'm now back with my first love, the Apple Macintosh. Performer, meanwhile, has continued to develop from a very basic but capable sequencer into one of the leading generation of 'MIDI+Audio' programs: Digital Performer. And so we meet again...
These days, the MIDI sequencer is fast becoming little more than a cut-down, lower-priced version of the digital music production tools that are now the flagships of the music software companies. The integration of audio tracks with MIDI tracks is changing the way that music is made, and the way that people think about sequencers. Instead of being the cheat's method of producing music with MIDI, utilising a computer to edit both MIDI and audio data has become accepted as a viable alternative to working with tape.
I've already commented at length about the way that my MIDI-focused, audio-free way of working was turned on its head by Opcode's Studio Vision Pro (see SOS March 1996), but despite changing my thinking about how you can use MIDI and audio in an integrated environment, I've yet to get around to investing personally in the requisite hardware for digital audio. Perhaps with this in mind, Mark Of The Unicorn approached me and offered the chance of a detailed look at Version 1.7 of Digital Performer. As a result, I've spent the last month working with a second example of a state-of-the-art combined MIDI and digital audio sequencer -- and so what follows is part review, part comparison with Studio Vision Pro, and partly a 'try this!' exploration of what it can do.
There are several ways of interfacing a Macintosh to MIDI: the simplest being the standard hardware interface, with Apple's MIDI Manager being the oldest and most basic software. But at the leading edge there are two options: Opcode's Open MIDI System (OMS) software and their 'Studio' series of hardware interfaces; and MOTU's FreeMIDI software and 'Time Piece' range of hardware interfaces. In the past I've always used OMS, so this was a chance to see how FreeMIDI behaved in practice.
As it happens, installing FreeMIDI was very smooth -- the automatic configuration detected the standard MIDI interface hardware that I was using. The current FreeMIDI (version 1.2.4) will only read pre-OMS 2.0 studio setup files, and so it couldn't read my present settings, but configuring a simple test setup with a few MIDI devices took only a few clicks of the mouse. The MOTU manuals are very detailed about the options available for installing the MIDI software and hardware, and having both FreeMIDI and OMS present in my computer's System caused no problems. Overall, getting the MIDI working was much easier than I expected.
MOTU have taken a conscious decision to support the higher end of the direct-to-disk/digital audio market, and so Digital Performer v1.7 does not support the rather limited audio facilities (and nasty connectors!) offered by Apple's Sound Manager 3.0 and upwards, nor does it support the Yamaha CBX hardware that the special version 1.41Y did. Instead, MOTU now concentrate exclusively on Digidesign hardware, from the simple plug-in cards of Audiomedia II (NuBus) and III (PCI), via what was once Session 8 and is now called Pro Tools Project, through to the fully professional and comprehensibly expandable Pro Tools III systems. Supporting just one manufacturer's equipment pays dividends in consistency and simplicity, and makes setting up much easier for the end user.
The review setup was based upon a low-end Pro Tools system, giving 4-voice polyphony and eight audio inputs/outputs. (There is more information on Digidesign systems in this month's Apple Notes column; see pages 122/3.)
Digital Performer builds music out of one or more phrases or Sequences. A Sequence can hold multiple tracks of MIDI and/or audio, and the basic window is called the Tracks window, where you can control the track settings (Instrument, Patch, Record, Play etc) and see an overview of the contents of each track -- blocks appear in boxes which represent beats, bars or sets of bars, and the darkness of the block reflects the density of the MIDI data. This lets you quickly see where the music is on each track. For more detailed editing, event or piano-roll style graphic windows can be opened, when desired.
The Chunks window contains all of the bits of music (the Songs, Sub-songs and Sequences) that have been created in step time or recorded in real time. Sequences can be dragged from the Chunks window to a Song window, so four separate 1-bar Sequences might be used to produce a 'chorus' Sub-song, and then this Sub-song could be dragged into another Song window which deals with the 'verse/chorus' structure of the final piece of music. The Song windows show the Sequences and Songs as graphic blocks, and this makes the structure of the song very clear -- probably one of the clearest ways of showing how a song is put together (and sadly missing in Studio Vision Pro!).
Personally, one of the most important aspects of any sequencer is how visible it makes the structure of the contents of the song. I tend to work with short sequences of between one and four bars, using them to build up my song in short sections before chaining these together into the final completed piece of music. Digital Performer makes this process very easy, and the hierarchical way that Sub-songs can be embedded within other Songs is superb.
MOTU have always used the graphics capabilities of the MacOS to the full. The user interface has evolved considerably over the years into a sophisticated, rich environment which is optimised for music sequencing. The most fundamental change to my eyes is that the familiar striped bar at the top of a window has gone. Instead there is what looks like a curved metallic bar with a cluster of icons at the left-hand end, and a set of steps leading to the name of the window on the right-hand side.
The icons have the usual 'close' box for putting a window away, except that it now has a triangle in it. The next icon provides a pop-up menu whose contents are specific to that type of window -- this saves having to move back up to the top of the screen for common functions, and is especially useful with larger screens. The 'move to back' icon is an arrow, and this moves the window behind all the other windows on the screen -- to the bottom of the pile! The minimise/maximise box is the final icon in this group of four which are common to all the windows, and although this location is not where you expect to find this useful tool (it is normally at the top right of the window), it makes a lot of sense where you have lots of differently sized windows on the screen. Because these four icons are always on the left-hand side, making a window fill the screen is easy because you do not have to find the right-hand side, you just find the much larger block of icons. Any additional icons after these first four are window-specific, and allow the setting of audio playback or soloing of a track.
In the same way, having the name of the window on the far right also makes things easier, because the name acts like a tag, and it is much easier to look along the top right-hand edges of windows than trying to find the middle of each top bar of a window. I can't help comparing Digital Performer's sequencer-specific user interface with Opcode's Studio Vision Pro. Digital Performer has a more 3D look, and the extra icons in the top bar of the windows make
it much easier to work with multiple windows on the screens at once, or make changing the mode of working quick and obvious. Conversely, Studio Vision Pro makes much more use of keyboard modifier keys to alter the effect of mouse clicks. Whilst this can be faster, once learned, it is not as intuitive as an icon that indicates what mode you are in.
My only criticism of Digital Performer's user interface is the slight lack of mouse-awareness. I like to use the mouse for everything, and Digital Performer does not allow some values to be changed by clicking and dragging. When changing a name, for instance, you can't use a mouse click outside of the box to finish typing -- you have to press the Return key.
Audio tracks appear in the Tracks window, and there are separate windows for viewing the audio soundbites -- the MIDI tracks can't be seen in the soundbite windows. The only concession is that you can see the bar lines, and so it is possible to align a graphic editing MIDI window and a soundbite window, though it is not as straightforward as opening a single window containing MIDI and audio tracks (as happens in Studio Vision Pro).
Soundbites can be dragged from the Soundbites window onto the Track window in much the same way as chunks can be dragged from the Chunks window into Song windows. In fact, the audio editing is mostly so intuitive (you drag soundbites around to move them in time, for example) that you don't need to read the manuals very much -- although there's plenty of very clear explanation in the 1200 pages of the four manuals. Audio editing is restricted to cutting, pasting, mixing, splitting (extracting part of a soundbite) and other tape-like functions -- if you need to make changes to the audio waveform, then you will need a separate waveform editor such as Digidesign's Sound Designer II. Mixer windows that look like the real-world equivalent can be used to control MIDI or audio tracks, and this is where you will find any third-party TDM effects plug-ins (if you have any installed).
Track windows do not indicate the actual length of a soundbite, they only show the first box that is occupied. This means that you need to check whether a specific part of the track has audio data in it; you can't see this directly from the Tracks window, so some of the overview capability is lost. This also makes editing audio and MIDI information more difficult, because although you need to 'Split' an audio track so that it forms a separate soundbite within the region, this is again not obvious from the Tracks window. This audio segment display problem will, I am told, be fixed in the next release of Digital Performer. In fact, it appears that the whole of the Tracks window segment display may change from one where the darkness of the segment blocks shows the density of the MIDI or audio data, to one where a tiny representation of the piano-roll graphic window is employed instead.
One of the most noteworthy aspects of Digital Performer has to be the innovative pitch-shifting and time-stetching facilities. Using native PowerPC code (the Digidesign DSPs are not used), the PureDSP processing firstly analyses any audio that is recorded and then allows some powerful changes to be made on anything that is monophonic. The analysis happens in the background, so you can carry on performing most operations whilst the computer is doing the analysis, and you can select a special display which reports on its progress.
Although this sounds like magic, the PureDSP processing does in fact work out the fundamental frequency of each sound, and then looks at the levels of the harmonics of that frequency (which actually determine the timbre). The analysis process thus splits the audio into pitch information and information about the harmonic structure of the sound (known as the 'formants'). The pitch extraction works best for monophonic sounds -- for polyphonic sounds Digital Performer provides conventional pitch-shifting and time-scaling.
Unsurprisingly, it does take time to perform this analysis: for example, a 30-second stereo soundbite (1,009,985 samples at 44.1kHz) took 200 seconds to analyse on my review machine, a Macintosh 7100/80AV, but subsequent pitch transpositions or time-scalings took only a few seconds.
Once analysed, it is then possible to manipulate the pitch, time or formants of the audio soundbite separately, and with astonishing quality. There are separate pitch transpose and time-scaling commands available within the program, plus the Spectral Effects control which lets you independently change pitch, time or formants as you wish. Extreme changes using PureDSP still affect the audio quality, but the usable range is far in excess of conventional processing.
With conventional pitch-shifting, where the audio waveform is merely replayed at a higher speed or chopped, the formants change if the pitch is changed. This sounds as if the person or instrument is growing bigger or smaller. In contrast, Digital Performer's PureDSP pitch processing changes only the pitched part of the sound and then reimposes the original harmonic structure. So the basic timbre of the audio is left virtually untouched, and the person or instrument appears to stay the same size, only the pitch changes.
One of the really clever features of Digital Performer is the way that the audio editing is integrated with the MIDI editing. Almost all of the editing functions work on a region basis: you select a region first, and then alter it with a menu command. For transposition or time-shifting, it is perfectly okay to select a region which contains both MIDI and audio tracks, and then to process both sets of information with a menu command. It is possible to set the audio transpose to be either the PureDSP transpose or the normal transpose, rather like the 'do not transpose' feature that you can set for drum tracks, and so a single transpose command on a specified region can behave differently depending on the content of the tracks -- but all that you do is 'transpose' the music. Brilliant.
This direct linking of audio and MIDI is particularly useful when you need to change the tempo of MIDI tracks, because it is easy to then make the audio tracks follow the changes. There can be quite a lot of background processing going on if you have lots of tempo changes, and this gives you some idea of how hard it would be to do this manually.
A lot can happen in 10 years: just about the only bit of Performer that I recognised since my last review was the pie-graph memory usage window! The improvements, however, are manifold and very impressive. For example, I actually prefer the time-independent, column-based display of song chunks to the more rigid, time-based display that I get from Studio Vision Pro. MOTU have clearly spent a lot of time researching how to make working with a sequencer easier -- the window management features may seem like superflous extras at first glance, but when you find yourself trying to push a window to the back in another program, then you know that they are actually very useful indeed.
Working in Audio windows is intuitive -- you drag soundbites around (even from the Soundbite window to the Audio window) to move them in time or from one audio voice to another. You have to select an audio sample by double-clicking on it before you can select parts of it for splitting (extracting the selected section) and this 'select and then edit' approach is consistent with the way that Digital Performer works -- you are always clearly shown which mode you are in -- and in this case, the double-click selection of the audio soundbite causes it to be surrounded by a box, so you then know that you are working on that specific soundbite. Holding the mouse key down to play an audio sample was fine once I got used to it, but I kept forgetting to turn off this option in the icon bar for the window, and so each time I tried to move an audio sample, it would pause and play before I could move it. Most of these minor irritations would, I feel sure, disappear once I became really familiar with the program.
Effort expended in learning to use the Audio window is easily offset by the sheer power of the audio processing. The PureDSP facilities make large pitch changes to audio tracks entirely feasible, without the accompanying chipmunk or creaky door syndrome. And the background processing of the audio means that you rarely notice it -- there were only one or two occasions when I did have to wait a few seconds; typically when changing a tempo map.
Digital Performer is obviously a MIDI plus digital audio sequencing environment for the user who requires comprehensive and straightforward editing and assembling of music, speech or sound effects. It does not have all of the more esoteric features that appeal to MIDI hackers and audio wranglers like myself, but then it isn't aimed at that market. Having said that, I wish the song chunking, window bottoming and a few other features of Digital Performer were present in the sequencer I currently use -- so I'm weakening.
Perhaps the best advice I can give is probably the least obvious: if you've invested significantly in digital audio hardware and a competing MIDI+Audio sequencer, then the cost of adding Digital Performer's very clever PureDSP processing, neat windowing and ultra-clear song chunk display is a fraction of what you have already spent on hardware -- so why not have two sequencers and use each for what they do best?
The 'PureDSP' processing of Digital Performer v1.7 has many uses. Here are a few of the discoveries that I made whilst I was exploring the less obvious corners of the program.
Because the pitch transposing thankfully avoids the normal 'munchkinisation' effect, large amounts of transpose can be useful for processing things you normally wouldn't attempt. Transposing speech up by an octave produces a remarkable 'Australian' twang. Shifting just the formants up by 12 half-steps produces 'cartoon duck' speech, whilst taking the formants and the pitch down by 12 half-steps gives a wonderful 'Second World War radio' sound. Although the audio quality suffers slightly for these extreme changes, it is very usable as a special effect.
Vocals are often the subject of extensive processing, and by deliberately misusing the time-scaling function, it is possible to create some very unusual and expensive-sounding effects. For example, by compressing the time by 1:2, and then expanding by 2:1, and perhaps even by 2:1 again, sibilants take on many of the qualities of the 'laser breath' effect.
When you 'split' part of an audio soundbite from a larger soundbite, the volume of the extract can often be too low -- and there's no obvious way to 'normalise' the volume so that it uses the full dynamic range. In fact, the 'Normalise' function is hidden away in the Mix menu option -- you merely select one audio soundbite and replace it with itself, normalising it at the same time.
The time-scaling, pitch transpositions and spectral effects processing can radically transform existing audio material. I took a drum loop off a CD, trimmed it and time-scaled its length so that it fitted exactly into one bar, and then extracted one drum sound out of it. By halving the pitch and using the on-screen 'draw the pan' feature, I had an unusual drum that jumped from one side of the stereo image to the other. I then mixed this into the drum loop and looped the drum loop to provide a backing beat. The part of the bar after the inserted drum sound needed some contrast now, and so I doubled the pitch of one of the subsequent drum hits and mixed that back into the loop. The result sounded like the product of a couple of samplers and lots of detailed editing, but it only took me a couple of minutes!
Digital Performer requires additional hardware support to enable it to work with digital audio, regardless of the Macintosh (or Mac clone) that you use. Currently, only the following Digidesign hardware is supported: HARDWARE OUTPUTS Audiomedia card (I, II, III, LC) 2 Sound Tools II 4 Pro Tools 4 Session 8 8 Pro Tools Project 8 Pro Tools III 16+ Here are some rough guide prices for Digidesign systems: A Digidesign, Avid Technology Europe Ltd, Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH. T 01753 653322. F 01753 654999.
Audiomedia II card £469 (includes Sound Designer II)
Audiomedia III card (PCI) £704.
Pro Tools Project (Session 8) £2,231 (NuBus or PCI).
Pro Tools III (16 channels) £6,238 (NuBus)
Pro Tools III (16 channels) £7,131 (PCI +extras).
882 Audio I/O £892.
888 Audio I/O £2,678.
Digital Performer requires additional hardware support to enable it to work with digital audio, regardless of the Macintosh (or Mac clone) that you use. Currently, only the following Digidesign hardware is supported:
Audiomedia card (I, II, III, LC)
Sound Tools II
Pro Tools Project
Pro Tools III
Here are some rough guide prices for Digidesign systems:
A Digidesign, Avid Technology Europe Ltd, Pinewood Studios, Pinewood Road, Iver Heath, Bucks SL0 0NH.
T 01753 653322.
F 01753 654999.
Digital Performer v1.7 requires a computer running MacOS 7.0 or higher, with 32-bit addressing enabled and at least 16 Megabytes of RAM. For computers running System 7.5 or higher, then 20Mb is recommended. Stripping unnecessary fonts, sounds, control panels, extensions and other items from your System Folder will probably improve stability and performance too. The Power Mac 7100/80AV used for this review had 32Mb of RAM fitted.
Computers with PowerPC chips will run the DSP code faster since it is 100% PowerPC native, although it will still run on 680n0 Macintoshes (eg. Quadra 650). Before buying a computer for use with Digital Performer, you should confirm its compatibility and suitability with Digidesign, because their digital audio hardware has more exacting computer requirements than MOTU's software.
The PureDSP pitch-shifting and time-scaling functions are superb.
Spectral Effects offer some novel changes to vocals and other instruments.
The DSP processing is well suited to a hard disk recording environment alongside MIDI tracks.
Audio processing occurs in background.
The PureDSP transposition is best for monophonic audio.
Apple Sound Manager not supported.
No Audio-to-MIDI or MIDI-to-Audio conversion.
For the sophisticated user who wants an audio software workstation for editing music, speech or sound effects, Digital Performer on a Power Mac will not disappoint.
£ £599 inc VAT.
A MusicTrack, PO Box 4, Arlesey, Bedfordshire SG15 6AA.
T 01462 733310.
F 01462 733390.