Paul White's absolute beginners' guide to MIDI continues with a look at MIDI's role in sequencing.
This month, he explains the concept of MIDI sequencing.To anyone used to playing and recording using traditional methods and skills, the MIDI sequencer is sometimes viewed as little short of cheating, but to the sequencer user, MIDI and sequencing are seen as practical tools that make complex multi‑part composition and performance a reality. Before MIDI appeared, few people could compose a symphony (or pop song, for that matter) and ever expect to hear it performed; now almost anyone can turn their musical ideas into a performance using affordable technology.
Before exploring the mechanics of sequencing, however, I'd like to tackle the idea that sequencing is somehow 'cheating' by looking at how things were done before the introduction of MIDI.
Having never personally written a symphony, I can't detail the exact stages involved, but I expect it goes something like this...
The composer sits at his or her chosen instrument testing musical ideas, and the ones that make it are then written down on manuscript paper for the various sections of the orchestra to play. The composer visualises (or should it be auralises?) the parts already written down while adding new sections, harmonies and so forth. Then, when the music is nominally finished, the score will be scrutinised and any required alterations or adjustments will be made.
Once the score is complete, an orchestra will be hired and given copies of the score, and the music will be played back as written by the composer. The composer, who may not even be able to perform to an acceptable standard on even one instrument, has conceived a piece of music and then written a list of instructions in the form of a musical score in order that a musically proficient orchestra can perform it. But has anyone ever accused Stravinski or Beethoven of being cheats, because they couldn't play all the orchestral instruments themselves? I think not.
In contrast, let's see how the MIDI composer writes. As with the orchestral composer, the work usually starts at the keyboard, but this time the keyboard is a MIDI instrument connected to a MIDI sequencer. Instead of writing down a score, the composer will record sections of the music into the sequencer against an electronic metronome set to the desired tempo. Instead of scanning a score to verify what's been done, it's a simple matter to play back the MIDI sequence to hear exactly what has been recorded. Best of all, you don't have to hire in an orchestra — a relatively inexpensive multitimbral synthesizer will provide all the sounds for you; each 'part' of the multitimbral synth plays one line of your electronic score.
In some ways, the sequencer is better than the written score, because it can play back a part exactly as you played it in the first place — it doesn't necessarily have to 'quantise' everything to equal subdivisions of a musical bar, as the written score does. And, just like the written score, if you're unhappy with something you've done, you don't have to start from scratch; you just erase the unwanted notes and 'write' in new ones.
When you summarise the way a musician composes using a sequencer, it isn't really too different from the way a traditional composer works. Both types of composer are likely to edit their compositions to some degree before they're entirely happy with them, and both bring in performers to play the finished composition. It doesn't really matter whether the finished piece is played by a bank of synths or by a hired orchestra whose role is simply to reproduce the composer's original work as faithfully as possible. My verdict, then is that electronic composition is as legitimate as any other form of composition. Note that I have no intention of fuelling the 'synths versus real instruments' debate at this point. If you have the talent to write a major symphonic work using synths, you can always get your computer to print out the score and have a real orchestra play it for you later!
Having covered the philosophical groundwork, it's now time to look more closely at the MIDI sequencer.
What exactly is a sequencer? It's often convenient to visualise a sequencer as being analogous to a multitrack tape recorder, and indeed, the 'layers' or parts of a sequence are recorded onto tracks, but it is vitally important to understand that what is being recorded is not the sound itself, but the electronic equivalent of a musical score. Just as a musical score is a series of instructions to the musicians, a MIDI sequence holds a series of instructions which tell your synths what to play. In some ways, a better analogy might be the player piano or pianola, where a punched paper roll holds the instructions that make the piano play, except in the case of MIDI, you have a multitrack, a virtual 'paper roll' capable of controlling many instruments at the same time.
In a typical setup, a MIDI instrument (usually, but not invariably, a keyboard) is connected to a sequencer via a MIDI cable, and when the sequencer is set to record, any notes played on the keyboard are recorded as MIDI data into whichever sequencer track has been selected for recording. In a simple system, you might have 16 MIDI tracks set up so that each is on a different MIDI channel, and if you feed the MIDI output of the sequencer to a 16‑part multitimbral sound module, you can play back all 16 tracks at once. If you only have an 8‑part multitimbral module, then you can only play back eight different sounds at once, in the same way that a real‑life string quartet can only play four lines of music at the same time.
To avoid having to switch the MIDI send channel on your keyboard every time you want to record onto a new sequencer track, modern sequencers convert the incoming MIDI data to the appropriate channel for the track you're recording on. This makes life very easy, because once you've completed one track, all you need do is select the next one and carry on playing.
The remaining capabilities of a MIDI sequencer bear more resemblance to a word processor than anything else. Like a word processor, you can delete or replace wrong characters (in this case, musical notes) and if you want to use the same phrase more than once, you can copy it and paste copies into new locations to save having to do the same thing lots of times. For example, if a song has the same structure for each chorus, you only need play the chorus once, then copy it to any place in the sequence where you'd like another chorus to appear.
Of course, there's more to MIDI data than notes, and a sequencer will record just about any MIDI data you throw at it, with the exception of MIDI clock — a sequencer has its own timing clock. Nevertheless, you can synchronise a sequencer to an external source if you wish, such as a tape machine (via a suitable sync box) or to a MIDI drum machine.
Unless you deliberately filter out certain types of MIDI data, you'll find that your sequencer captures Note On/Off, Pitch, Velocity, Aftertouch and Controller information as well as MIDI Program Changes and even System Exclusive (SysEx) data. If these terms are unfamiliar, fear not — we'll be looking at Controllers and Program Change information next month. There's even less need to worry about the concept of SysEx data at this point, but it is useful to know that it is possible to record a SysEx dump of all your synth sounds at the start of a song, so that when you first play the sequence, your synths are automatically loaded up with the appropriate set of new sounds to play that particular musical sequence.
A MIDI Program Change command recorded during the count‑in period of a track will ensure that the connected synth switches to the correct sound patch before playback commences, but you can also insert Program Changes part way through a track (as many times as you like) if you want the sound to change for, say, a solo. This is the orchestral equivalent of writing a note on the score at a certain bar number to tell a violin player to put down his violin and play the next part on a flute! This isn't something you'd usually do in real life, but a MIDI sound module is equally proficient on all instruments and, as yet, MIDI modules don't have trade unions!
When your sequence is played back, the sequencer transmits the MIDI information to the receiving synth(s) — or sampler, drum machine, and so on — in exactly the same order, and with the same timing as you originally played it. If you so wish, you can change the tempo after recording without affecting the pitch (unlike a tape recorder, where you're dealing with sound rather than MIDI data). If you're still not sure why the pitch doesn't increase as the tempo goes up, think back to the orchestra and score analogy; if the conductor asks for a piece to be played faster, the orchestral instruments don't change in pitch. Similarly, if you pedal a pianola faster, the paper roll will be played faster but the piano's tuning will remain the same.
In reality, MIDI does has a finite timing resolution, because the sequencer or computer sending the MIDI information has to work to an internal timing routine based on an electronic clock. However, in practice, MIDI is far more accurate than a typical human performer, and is capable of resolving a bar of music into at least 960 time divisions, and frequently more.
All MIDI sequencers are based on computer technology, but you have a choice of buying a sequencer system that runs on an existing computer (such as an Atari ST, Apple Mac, Apple Power Mac, IBM‑compatible PC or Commodore Amiga) or opting for a piece of dedicated hardware where everything you need is built into one box. The two types work in a similar manner — what tends to vary is the way in which the recorded information is displayed, and how easily it can be edited.
For those who are relatively accomplished players, hardware sequencers (like the Roland MC500) offer the benefits of simplicity and convenience, but they rarely have the information display capability of a full‑size computer screen. And because there's no computer mouse, editing is generally less comprehensive and more time‑consuming than it would be on a computer‑based system. However, the recording process is usually just a matter of hitting the Record button and playing. A significant benefit of hardware sequencers is that they are more practical in live performance situations; they are more compact and more rugged than a computer‑plus‑monitor, and you have fewer things to plug in.
Has anyone ever accused Stravinski or Beethoven of being cheats, because they couldn't play all the orchestral instruments themselves? I think not.
Some MIDI sequencers (including all the computer‑based ones) lose all information stored in their memory (RAM) when they are switched off, so it is vital to save your work to disk at regular intervals. The MIDI information which makes up a completed song can normally be saved onto either floppy or hard disk in the form of a song data file, and a single floppy disk will hold several songs of average complexity. Many hardware sequencers also have a built‑in disk drive, allowing songs to be saved as files on floppy disk, though some of the less expensive models (such as the Alesis MMT8) use battery‑backed‑up memory instead of disks. Once the memory is full, however, you either have to save your work to a MIDI data filer (which has an in‑built disk drive) or throw away your old project before you can start a new one. Usually, this kind of sequencer can only store a few songs at a time.
The computer‑based sequencer is capable of more sophistication than most hardware models, which means there may be a steeper learning curve. However, this is more than made up for in my opinion by the amount of visual feedback available, especially when it comes to tasks like song arrangement.
Unless you're using an Atari ST, which has a built‑in MIDI connections, you'll have to buy an external MIDI interface box, though some of the newer GM synth modules (such as Yamaha's MU50) come with PC and Mac MIDI interfaces built in. MIDI interfaces for Apple Macintosh computers usually plug into the modem or printer ports on the machine, while PC users need an interface card which goes inside the computer. A basic sequencing setup is shown in Figure 1, and to keep things simple, I've depicted a 'dumb' master MIDI keyboard; if you have a MIDI keyboard that includes a sound generation section, simply select Local Off and connect it up like any other synth module.
The majority of the leading software sequencing packages have adopted the style of user interface pioneered by Steinberg in their Cubase software. This typically comprises a main screen page, which handles the basic 'recording' and arranging, plus a number of further pages which address various aspects of editing and, where applicable, scoring. The record and playback controls are designed to look something like a tape recorder's transport control, and the edit pages usually allow you to examine (and change) the recorded data in several ways: (i) as a list of MIDI events; (ii) graphically, in the guise of a 'piano roll' display; or (iii) in the case of 'score' versions, in the form of a conventional musical score.
Some software sequencers include sophisticated scorewriting facilities which enable you to print out sheet music for your compositions, in which case you'll need a printer that is compatible both with your computer and the software package. However, some musical literacy is useful, because the computer doesn't always interpret what you play in the same way that a trained scorewriter would.
From the editing pages of a typical sequencer, you can change the value, timing, and velocity of any of the notes you've played. Alternatively, you can build up compositions manually, by placing new notes onto the quantise grid in non‑real‑time, rather like writing out a manuscript. The non‑real‑time entry of note information may also be referred to as step‑time entry.
A number of related non‑destructive (ie. the operation is not permanent and can be reversed) editing options are sometimes available, including the ability to transpose your music, either as you play or after recording. You also usually have the ability to make the music louder or softer by adjusting the overall velocity. On some systems, you can even compress the dynamic range of your MIDI data to even out the difference between your loudest notes and the quietest ones, as well as delay or advance tracks relative to each other (to make timing adjustments). This is frequently achieved by recalculating the note data during playback, but the real data isn't changed, so you can always revert to your original performance data.
It is possible to sequence the sounds from your drum machine just as you can any other type of MIDI sound module, but remember to turn off the drum machine's external MIDI sync first, otherwise every time you start your sequencer, the drum machine's internal patterns will start to play. Unlike a conventional instrument, where each note on the keyboard plays a different pitch of the same sound, drum machines place different sounds on different keys, allowing access to many varied drum sounds. Because it's difficult to play a complete drum part in one go via a keyboard, it is common practice to spread the drum part over several sequencer tracks — enabling you to record, say, your bass and snare first, your hi‑hats next and finally your fills. This method of working makes it easy to edit your drum tracks later, without having to work out what note corresponds to what drum sound. And once the drum part is completed, of course, you can always merge the drum tracks into one for convenience. Most sequencers offer a suitable track merge function these days.
MIDI sequencers are very powerful tools both for music composition and recording, and because they have grown so sophisticated, there are still a great many features that I haven't discussed. For example, MIDI allows you to remotely control the volume of your instruments, so by recording MIDI Volume information (Controller 7) in your sequences, you can create automated mixes.
Wonderful though sequencers are, they are still far from perfect. Aside from the inevitable software bugs that creep in, they tend to force you to work in a way that you probably wouldn't adopt if you were playing and composing conventionally. Most insidious is the metronome or tempo click that you have to play along to, and although you can turn this off and record 'freestyle' regardless of bar positions, you won't be able to quantise your data (for an explanation of quantisation, see the box elsewhere in this article), and you won't be able to print out a meaningful score. This means that tempo changes have to be planned ahead rather than being intuitive. Although software designers are now including features to help you in this area (such as re‑barring), it takes a lot of determination to move away from the fixed tempo, four‑to‑the‑bar, music that we've all become so accustomed to.
Despite the pitfalls mentioned, MIDI sequencing still offers far more advantages than disadvantages, and used creatively, it makes many things possible that would have been far too impractical or expensive in the pre‑MIDI era. And finally, don't think that sequencing is difficult — once you've made a start and seen how easy it is to handle the basics, you'll wonder why the manuals ever needed to be so thick!
MIDI sync was covered in some depth in Part 1 of this series, but it is useful to recap here on the main points.
Sequencers with integral hard disk recording facilities offer a great way of combining audio with MIDI, but they still tend to be expensive and there's also the problem of backing up very large audio data files. Because of this, most people still use multitrack tape, but there's no advantage in recording your sequenced material to tape if you can find a way of making the sequencer run in sync with your multitrack.
The easy answer is to record some form of MIDI sync code onto tape. This means you give up one tape track to record the necessary sync code, but you gain as many 'virtual tracks' as your sequencer and synth/sound module collection can provide. The simplest way to achieve this is to use a 'Smart FSK' MIDI‑to‑tape sync box which you can buy for as little as £100. These use both MIDI Clock and MIDI Song Position Pointers to ensure that your sequencer starts at the right time and remains in perfect synchronisation with your multitrack, regardless of whether you play the tape from the start of the song or from half way through. For more on MIDI sync, FSK, and Song Position Pointers, take a look at the article 'Synchronisation Explained', starting on page 186 of July '94's SOS.
What basic features can you expect from a MIDI sequencer? Obviously every sequencer is different, but all should be capable of the following core functions.
- REAL‑TIME RECORDING:
You play in your MIDI data from a keyboard and record just as you would with a tape recorder. Unlike a tape recorder, you can transpose, change tempo and quantise your data after recording. If you want to use the quantise feature, you have to play to the internal metronome track when recording.
- STEP‑TIME RECORDING:
Notes are played in one at a time — it's rather like typing a letter with one finger! You decide where the notes go and how long they're going to be, after which you can play back your work at any tempo. Most people mainly use real‑time recording with occasional recourse to step‑time when the going gets tough. With a piano‑roll type of editing screen, you can also 'draw' your notes directly onto the quantise grid and then use the mouse pointer to 'stretch' them to the desired note length.
A typical sequencer will have a variety of editing tools to enable you to change your composition once you have recorded it. These range from the ability to change individual notes to the ability to change entire arrangements and swap instruments.
This is the ability to move your notes to the nearest accurate subdivision of a bar (for example, 16ths of a bar). See 'Quantisation' box elsewhere in this article for more details.
Notes can be transposed by any amount without altering their lengths, while entire compositions or sections of compositions can easily be shifted to a different key.
Any section of music can be copied to different tracks or to different locations in the song. This is useful for duplicating repeated sections, such as a chorus, or for doubling up a line of music on two tracks by assigning them to different instrument sounds. Cut allows unwanted material to be removed.
This may be implemented in a separate part of the program, as in C‑Lab's Notator/Creator, or it may be handled using Cut/Copy/Paste as in the case of Steinberg's Cubase, EMAGIC's Logic and Opcode's Vision.
Most sequencers allow you to mute tracks, or solo a track, so that you can hear it in isolation.
Simply allows you to continually loop around a specific section of music while you record or edit.
Computer‑based sequencers, and some hardware sequencers, have an Undo function which, as the name implies, lets you undo the last seemingly permanent thing you did.
One important feature common to both hardware and software sequencers is the ability to quantise data after recording — a useful feature for those musicians not possessed of a perfect sense of timing. Essentially, when you choose to quantise something, the timing is changed so as to push each note you've recorded to the nearest exact subdivision of a bar. For example, if you are working in 4/4 time, and you select 16 as your quantise value, all the notes becomes locked to an invisible grid which in effect divides the bar into 16 equal time slots. Quantise must be used carefully as it can strip all the feel from some types of music; however, if you're doing dance music where precise timing is essential, the quantise feature is indispensible.
The most recent computer‑based sequencers allow you to unquantise data as well as to quantise it, but be aware that some less advanced software sequencers and a number of hardware sequencers perform what is known as destructive quantise. So if you think you might need to go back to the original version, it's vital that you keep a copy of the sequence.
Another feature which I find really valuable is what is usually referred to as percentage quantise. Using this, you don't have to make all your notes snap to the quantise grid; instead, by setting a quantise value of say 50%, you can have your notes moved to a position that's half way between where you originally played them and the nearest time slot in the quantise grid. This is great for tightening up your playing without losing all the feel.
Yet another quantise‑related function is swing, where the quantise grid is moved away from regular slots to alternating longer and shorter slots. This can be used subtly to add feel or used more aggressively to turn a 4/4 track into a 2/4 track, say. It's now even possible to load in third‑party groove templates (such as DNA Grooves) created from the timing of real players.
Software sequencers have several obvious advantages over hardware sequencers, but that doesn't mean that they're better — it all depends on what facilities you need and whether you want your sequencer to be portable.
The main pros of software sequencers are as follows:
- A good visual interface.
- More comprehensive editing facilities.
- You can still use the computer for other purposes.
- You're not tied to one manufacturer for software upgrades — if somebody comes out with a better program, you can always move over to it.
- Most computer sequencers support multiple MIDI output ports via a special multi‑port MIDI interface. This means you are not restricted to 16 MIDI channels and a typical system will provide six output ports, giving you up to 96 MIDI channels to work with.
- The most popular sequencer software packages now allow you to transfer song data from one computer platform to another and, in some cases, even from one manufacturer's software sequencer to another's.
- Professional standard score printing is available from many sequencing packages, using either an inexpensive ink‑jet printer or a laser printer.
Hardware sequencers have their advantages too, the main ones being listed below:
- One‑box solution to sequencing.
- Generally more reliable than computers in live situations or when being moved from studio to studio.
- Although they may have fewer editing options than a software sequencer, they also tend to be easier to use.
- You don't have to learn to use a computer before you can begin to learn your sequencer software.
A basic MIDI sequencing setup starts at your keyboard — it's here that the MIDI information to be recorded originates. The master keyboard is connected via its MIDI Out socket to the MIDI In of your MIDI interface, or directly to the MIDI In of your hardware sequencer or Atari ST. As mentioned earlier, if your keyboard includes a synth section (in other words, if it makes sounds), then turn Local Off and patch a MIDI cable from the sequencer's MIDI Out to the keyboard's MIDI In.
If you have other MIDI modules in the system, you can daisy‑chain them in any order by feeding the MIDI Thru of one piece of gear to the MIDI In of the next, as described last month. Up to three modules can normally be chained in this way without problems, but longer chains may cause stuck or missed notes (due to corruption of the MIDI signal), in which case you should use a multiple output MIDI Thru box connected to the output of your sequencer and then feed each module (or short chain of two or three modules) from separate outputs on this Thru box. MIDI Thru boxes were explained in Part 1 of this series, last month.
If you've connected up your system as described but no sound comes out, here are a few things you might want to check before you get into serious panic mode.
- Check that everything is switched on and that your synth modules are set to Multi or Sequencer mode (assuming you want to use them multitimbrally).
- Check your MIDI cable connections and don't rule out the possibility a faulty MIDI lead. Some modules have a combined MIDI Out/Thru socket; if so, ensure MIDI Thru is enabled (see handbook for that piece of equipment). To help narrow the problem down, most sequencers have some form of indication that they're receiving MIDI data and many modules have a MIDI light or other indicator that blinks when data is being received.
- Check that you've set the MIDI channels correctly on your modules and that Omni mode is switched off on all modules. If two or more instruments try to play the same part, the chances are you've either got more than one module set to the same MIDI channel or something's been left set to Omni. If your master keyboard plays its own sounds when you're trying to record using the sound of another module, check that Local Off is really set to Off.
- If playing a single note results in a burst of sound, rather like machine gun fire, or if you get stuck notes or apparently reduced polyphony, the chances are you have a MIDI loop. In a MIDI loop, MIDI data generated by the master keyboard passes through the sequencer and somehow finds its way back to the input of the master keyboard, where it starts its round trip all over again, rather like acoustic feedback. This usually happens when you are using a keyboard synth as your master keyboard and have forgotten to select Local Off.
If you have one of those rare instruments with no Local Off facility, you'll probably find that your sequencer allows you to disable the MIDI Thru on whatever channel your master keyboard is sending on (most people leave it set to channel 1).
If you are unfortunate enough to have neither facility, then all you can do is record with the MIDI In disconnected from your master keyboard and use the sounds from external modules. When you've finished recording, you can reconnect the master keyboard's MIDI In, if you wish, and use it to play back one of the recorded parts or to layer with an existing synth voice.