The third part of our series on MIDI fundamentals examines the challenges and possibilities offered by MIDI sequencers and looks at their place in a typical electronic music studio.

CONSIDERING THE NUMBER of references to MIDI in your favourite magazine, it's easy to get the impression that the whole musical world revolves around this four-letter acronym. In some ways, that's not far from the truth because so many people are using it to do so many different things. If you've been following this series, you've probably already spotted some of these possibilities yourself, but this month we're going to look at one of the most powerful and most important - but also most confusing - applications available: MIDI sequencing.

Before diving into an explanation of sequencing, let's spend a second or two reviewing the basic concept of MIDI. As pointed out in part one, MIDI is a hardware and software communication standard that allows various pieces of musical equipment to "talk" to each other. The "conversation" which occurs consists of messages sent over a MIDI cable that describe musical events - using the term "events" loosely. If you connect two instruments, as "master" and "slave" (as described in part two), then the messages sent from the master instrument will tell the slave instrument(s) exactly what to do. The result is various instruments working together as a unified system.

What's a Sequencer?

AT ITS MOST basic level, a sequencer can be thought of as an extra pair (or pairs) of hands which record and play back music. The programming need not be a tedious or confusing process though, it can be as simple as playing something on a MIDI keyboard. In fact, all you usually have to do with a sequencer is hit Record and play. After you've finished you hit Play and the sequencer will automatically play the keyboard, not unlike an old player piano (who remembers them?). Sequencers also usually allow you to enter notes one at a time, however, so you can also use them to create music that you couldn't otherwise play.

One of the greatest things about sequencers is that they can play many different things at once. Like tape recorders, most sequencers have a number of separate tracks for recording different parts. If you want each part to be played on a separate synthesiser, all you have to do is change MIDI channels before you record each part and the sequencer will remember the channel assignments along with the parts, thus keeping them separate. On playback the sequencer sends the information out to whatever instruments you have connected to it. So a sequencer forms the centrepiece of a MIDI music system; it sends MIDI data to each of the individual units and acts as its co-ordinator.

A sequencer can provide these capabilities because its primary function is to record MIDI data. In fact, that's why you may hear some sequencers referred to as MIDI Recorders. Beware though, sequencers do not record audio information, only MIDI messages of various types.

Sequencer Types

SEQUENCERS BASICALLY COME in two varieties: stand-alone hardware devices or "dedicated" sequencers (including the "built-in" variety like that found on the Ensoniq ESQ1), and software programs designed to run on various personal computers. Each of these has its advantages and disadvantages. Defining your own specific needs is the most important factor in determining which type to use. The main advantage of a dedicated sequencer is its portability and general ease of use, but these often lack the memory capacity and comprehensive editing facilities of software programs. Computer-based sequencers, on the other hand, require a MIDI interface for your computer to allow it to communicate with other MIDI equipment, but allow you to see your work on-screen and deal with a lot of information at once.

The other important difference between various types of sequencer - regardless of whether they are of the dedicated or software variety - is the kind of MIDI data they can record, and the extent to which you can edit that information. All sequencers can record basic Note On Channel Voice messages, but they can't all record MIDI controller information, such as aftertouch or sustain pedal, and other less-common MIDI messages. Similarly, most sequencers allow you to cut and copy sections of music, but not all of them permit you to manipulate the data at event level - like altering individual note velocities or pitch-bend messages.

Recording a Sequence

A BASIC SEQUENCING setup is shown in Figure 1. To record a sequence you need to select a MIDI channel on the keyboard, set the tempo on the sequencer and hit Record. The reason for setting the tempo is that a sequencer records the length of the notes as well as the pitches, and so it needs to have some sort of reference to determine how long those notes are. Most sequencers put out a click (metronome) at the tempo you've selected so you can play along with it.

Though it is their main function, sequencers can generally do more than just record and replay MIDI data; they often allow you to edit it, and that's where their real power lies. Whether you want to erase, copy, or transpose a section of music, or just fix the velocity level on that C# in bar 62, a sequencer con, depending on its specific features, provide that flexibility.

Sequencers can also be considered as storage devices for any musical ideas that you may have. Though the obvious analogy is the tape recorder, they actually turn MIDI data into a form of storage that contains a similar amount of detail as traditional music notation. Sequencers record the notes that are played, their volume level, their length, as well as initial parameters such as time signature and tempo. This similarity has resulted in the development of software programs which convert MIDI sequence data into standard Italian notation.

"Sequencers come in two varieties: 'dedicated' hardware sequencers and software programs designed to run on various personal computers."

The Messages

SEQUENCERS PRODUCE ANY type of MIDI messages which they are capable of recording because they simply play back the data that is recorded into their memory. In addition to the Channel Voice messages covered last month, sequencers can produce and respond to System messages such as System Real Time Clock and System Common Song Select and Song Position Pointer messages. (Considering that a sequencer often acts as a system controller this is rather appropriate - you see, this stuff really does make sense after all.)

Clock messages are used to synchronise two or more devices which require timing information, such as a drum machine and a sequencer. The MIDI Clock messages ensure that the two devices will play at the same tempo and the Start, Stop and Continue commands, which also fall under System Real Time messages, make sure that the instruments start and stop together. In short, they keep things in time and make sure everything starts and stops on cue - try getting your band to do that more than twice a night. The MIDI specification calls for 24 clock pulses (or clocks) to be sent per quarter note. This may be presented as 24ppqn. The speed at which these clock messages go out over MIDI determines the tempo at which the music will play.

Song Select messages basically function like Program Change commands for sequencers; they inform the receiving instrument - usually a drum machine or a sequencer - to call up a specific song number. (Most sequencers divide their memory into patterns and songs, with songs being an ordered collection of patterns.) Song Position Pointer messages, on the other hand, refer to a location within a song. Devices which implement this important type of MIDI message keep track of the number of sixteenth notes that have occurred since the beginning of the song and determine the location based upon this count. If a device which recognises song position pointer data receives such a message, it will position itself at the appropriate point in the currently selected song.

Some sequencers also offer the ability to record and remember the timbres used to play the music. Depending on how deep you want to get into this, as well as the capabilities of the sequencer you're using, you can use either Program Change commands or actually store the synth program parameters via System Exclusive (or SysEx) messages.

The System Setup

OK. NOW ITS time for some fun. To really understand how a sequencer works and why it's such a useful tool, you need some more specific practical examples. Starting from where we left our system last month, the master keyboard and expander module are connected via a MIDI switcher box. So the first thing that needs to be done is to connect the sequencer into the system (see Figure 2). In this example, the MIDI Out of the master keyboard is routed to the sequencer's MIDI In and vice versa, so the expander can be controlled by either of the two - depending on which input is switched in as the controller. Because the sequencer initially has no information to send to the expander, the original setup should have the keyboard controlling it.

Before you can begin recording with the sequencer, you need to decide which channel you want to record on. You also need to make sure your keyboard controller is in Omni Off (Mode 3 or 4) because with sequencers, independent channels become essential. Generally, you should choose the channel by determining the transmit channel of the keyboard - the sequencer will record that information as well as the notes you play - but some sequencers require that you enter a channel for each track. One interesting channel feature that some sequencers have is a "rechannelise" function, which allows them to alter the channel number of the information they receive. If you own a DX7, which can only transmit on MIDI channel 1, or if changing channels on your synth is a difficult process, this feature is a life-saver.

And speaking of channels, it's time to clear up the common misconception about the difference between tracks and channels. A channel refers to the MIDI channel that the notes or other MIDI information was recorded on, while a track is generally the location in memory where that information was stored. Hence a channel can be thought of as a subset of a track. For example, some sequencers have only two tracks - such as Yamaha's QX7 and QX21 - but each of these tracks can hold independent note information for each of the 16 MIDI channels.

"An important difference between sequencers is the kind of MIDI data they can record and the extent to which you can edit that information."

So unlike tape recorders, a two-track sequencer can hold multiple channels of musical data without any degradation whatsoever - remember, sequencers do not record audio. Instead of bouncing down, as you would on a multitrack tape recorder, sequencers with a limited number of tracks generally have a Merge function which allows you to combine separate information on different MIDI channels onto the same track. Some sophisticated sequencers also include an "unmerge" function which allows you to separate individual channels after they've been combined - a feature which people who work with audio recorders would kill for.

Other sequencers have huge numbers of tracks (would 256 be enough for you?), each of which can contain information on a single MIDI channel, if not on all 16 channels. Some of the reasons for having such a large number of tracks are: easy editing of the various independent lines (sequencer editing is often done by track), spreading dense MIDI data over a number of different channels (this helps prevent slow data transmission) and the ability to compare a number of different versions of a particular musical line, such as different takes of a solo.

Another point to be aware of is that some sequencers can simultaneously record on a number of MIDI channels. If you have a sophisticated master keyboard which can transmit on more than one MIDI channel or if you're dumping a sequence from one sequencer to another, this feature can save you a great deal of time.

The last point that has to be considered before you begin recording has to do with filters - not the kind found on your synth or in your coffee machine, but data filters. Many sequencers have the ability to selectively ignore certain incoming MIDI messages (to conserve memory) and you need to decide which messages you want to record and which you can afford to ignore. Continuous controllers such as aftertouch and pitch-bend, for example, eat up memory because they constantly send out location messages and these take up almost as much memory as a Note On message. Consequently, unless you really need to record these, you may find filtering them out very useful.

An alternative is to sequence a particular line without pitch-bend and then add that later on. You can do this in a couple of ways: first of all, you can record a separate track of pitch-bend information alone and then merge that with the notes you've already recorded - remember, sequencers record MIDI information, not necessarily just notes. The second way is to use the pitch-bend wheel as the sequence plays back through the synth, so that when you mix down onto audio tape you can add the pitch-bends (or modulation using the mod wheel) on the final recording without having to waste space in your sequencer. This may not be the most elegant or accurate solution, but it does work.

ONCE YOU'VE ARRANGED your system to your liking, you can begin sequencing. Pick a track, double check your mode and channel status on the master synth, hit Record and play something inspiring (or at least try to). Assuming that the master keyboard can only play one timbre at a time, but that the expander is multitimbral - that is, it can play a number of different patches at the same time by responding to information from a number of MIDI channels. (As soon as you start to do any sequencing, you'll recognise the importance of multitimbral operation.)

When you've finished recording, hit Play and you should hear an equivalent performance of what you played into the sequencer, though it will depend somewhat on the settings of features like quantisation and other things specific to individual products. You can adjust the tempo as much as you want and you won't affect the pitch because, I say again, sequencers record MIDI messages not audio signals.

Back to our system. Whilst you've been playing the sequence back you shouldn't have heard anything from the expander because its input is being controlled by the keyboard. Even if you change the setting on the MIDI switcher, the expander will only respond to the sequenced data if any of its receiving channels match up with the sequencer's output. If you want a sound from the expander to be layered on top of the sound from the master keyboard, then make sure the channel for that specific patch coincides with the channel for that note information, but otherwise keep them on different channels.

"Editing features on sequencers may not always be as flexible as a piece of paper, a pencil and an eraser, but they do offer interesting possibilities..."

To record the next track, the first using voices from the expander, you need to make a few small adjustments. First, decide which MIDI channel you want to record on and set it up as the transmit channel on the master keyboard and the receive channel on the sequencer - if your sequencer requires you to do this - and the expander; they all need to match up. Next, make sure the controlling input of the expander is set to the keyboard. Next, check to see that the receiving channel of the master synth is the same as the channel its track was originally recorded on - that is, if your synth can operate on different send and receive channels. (If not, I'll explain how you can overcome this problem.) Finally, turn the local control of your master keyboard off (so that it will not play its own internal sound circuitry) again, if you can.

The reason for these last two steps can be summed up in one word: monitoring. You'll probably want the sequencer to play back the first part you recorded on the master keyboard as you record this second track, but if the local control is on, you'll end up having both parts playing on the master synth. If you only have a limited number of notes available, you'll quickly run out.

Sequencer in the System

To record a third track, you need again to select the appropriate MIDI channels and then make one more small change: switch the expander's controlling input to the sequencer. If you try to play something now, however, you'll probably not hear anything because the MIDI data from the keyboard is going into the sequencer and staying there. Of course you could connect the sequencer's MIDI Thru to the expander's MIDI In, but then you wouldn't be able to hear the previously sequenced parts coming from the sequencer's MIDI Out.

So how can you hear both? Glad you asked. Most sequencers have a function called MIDI echo - or something along those lines - which basically combines the incoming MIDI data from the controller with outgoing sequenced data by "echoing" the incoming MIDI information into the rest of the system via the sequencer's MIDI Out port. In other words, a MIDI echo feature merges what would normally be MIDI Thru data with the MIDI Out information. (By the way, you could have set the expander to have been controlled by the sequencer from the start, but for purposes of the example, I wanted to explain the possible routings of the MIDI data.)

If your master keyboard can't transmit and receive on different channels and it doesn't have a Local Off feature, you can still sequence using the basic method I've described, but you need to make a few more changes. First of all, you have to change the order in which the tracks are recorded: the expander tracks need to be done at the beginning. By using this method, all you have to worry about is changing the transmit channel on the master keyboard for every different part you record. The receive channel's setting doesn't make any difference until you record the final part, using the voices from the master keyboard. The second thing you have to do is even easier: turn down the volume on the master keyboard to simulate the local off feature. Though there is a difference in the internal operations of the synthesiser, this will work just as well.

Editing

IF, FOR SOME reason, you're not happy with the masterpiece you've begun to create, there's still hope. As I've already pointed out, one of the most important aspects of sequencing is being able to go back and change the MIDI data so that the final result, the music, sounds the way you want it to. Editing features on sequencers may not always be as flexible as a piece of paper, a pencil and an eraser, but they do offer a number of interesting possibilities, including some things that would be impossible with recording tape. Specific editing features vary wildly from sequencer to sequencer, but it's certainly worthwhile to explore some of the options available.

On a very basic level, nearly every sequencer on the market allows you to transpose any music you've recorded into it. With the Cut, Copy and Paste functions found in many software programs - and some dedicated units - you may also be able to transpose just one small section of a piece and then add it somewhere into the existing sequence.

If your timing is not quite as good as it should be, most sequencers have a feature called quantisation, or auto-correct, to help out. Though a few different types are available, the most common form of quantisation moves the position of the beginning of a Note On message to a multiple of the chosen note value. In English that means if you select a quantise value of sixteenth notes, every note will be moved to a point where it coincides with a sixteenth measure of a bar. Most sequencers offer a number of different quantisation levels, so you can "correct" different parts of your music to different degrees. But beware - if it's overused, quantisation can rob a sequence of any human feel.

Another important feature with which you can fine tune your sequence is step editing, or step recording. Up until now, all the examples have used real-time recording, which is similar to recording with audio tape. You hit Record and play along with the click supplied by the sequencer. But sequencers can also record MIDI data one step at a time, so if your keyboard chops aren't up to scratch you can still take advantage of sequencers by inputting notes or chords one at a time. Step recording - which is often referred to as an editing function - differs from sequencer to sequencer but generally they only allow you to enter note data, including velocity levels, and some Channel Voice messages such as program changes.

With the various types of features available on different sequencers, most of them save their sequence data in a proprietary format. A recent addition to the MIDI specification promises to change all that, however. The MIDI File addendum outlines a standard format for sequencer files which will allow sequences recorded on one sequencer to be played back by and edited on a different one. In addition to saving time when transferring information between sequencers, this new type of System Exclusive data will permit you to take advantage of the best features in different sequencers. For example, you could record a sequence on a sequencer with a good input stage and then edit it on another which you feel has better editing features. The creation of MIDI Files may also lead to the development of powerful generic editing software programs and other interesting new developments.

Next month we'll bring rhythm into the picture by adding a drum machine to the system.

Until then, happy sequencing.