Two of the most important developments in music technology during the 20th century were digital audio and the Musical Instrument Digital Interface (MIDI). Those developments still have an ever-increasing impact on the ways that we create, perform, and listen to music and will undoubtedly transform music education in the 21st century. But when should you use digital audio in the classroom (or on your own recordings, for that matter), and when should you opt for using MIDI? The answer is “It depends.”

Fig. 1: Most sequencers allow you to view MIDI messages in several ways. In the Piano Roll view, MIDI tracks are commonly represented as colored blocks, while audio tracks are represented as waveforms.

Almost all modern sequencing programs, in addition to many other music programs, offer both MIDI and digital audio recording and playback (see Fig. 1). Some hardware electronic instruments, especially higher-end keyboards and sound modules, also provide a range of digital audio and MIDI features. Digital audio and MIDI are both digital datastreams, but that's where the similarities end. Once you understand the basic differences between the two technologies, you will be better prepared to decide which one is better suited for a particular application in the music classroom and when you should blend them.

The most important concept to grasp is that MIDI is not a form of audio. Rather, MIDI is a standardized protocol that allows MIDI-compliant hardware and software to communicate. Just as we communicate with each other using a common language such as English or Spanish, MIDI is the universal language that allows all MIDI-equipped devices and programs, regardless of manufacturer, to control or respond to other MIDI devices and programs. MIDI “conversations” typically consist of instructions called messages (see Fig. 2). MIDI-compliant products include effects processors, software, electronic musical instruments, sound modules, and even nonmusical devices such as banks of stage lights. The MIDI Specification is overseen by the MIDI Manufacturer's Association (MMA), at www.midi.org.

Perhaps the most common MIDI message is Note On (play a note), which is accompanied by a note number that indicates which note to play and a Velocity value that indicates how quickly the note was attacked. A Note Off message means “stop playing the note,” and again is accompanied by the note number. Other common messages are Volume (how loud to play) and Pan (whether the sound will go to the left or right stereo channel), and there are many more. MIDI is analogous to print music in that print music tells the musician when to play a note, what note to play, and when to stop playing it. MIDI does not generate sound in and of itself; it conveys to a device what it must do.

Let's say you record a MIDI track with a keyboard synthesizer and MIDI-recording software (a sequencer). The MIDI track isn't recording the sound of the synthesizer. Rather, it records something to the effect of “using an acoustic grand piano sound (Program 1) and playing at a tempo of 120 bpm, first trigger Note Number 60 (middle C) by hitting the key moderately hard (say, a Velocity value of 70). Hold it for a full beat, then release it (Note Off). Next, simultaneously trigger the notes D (62), F (65), and A (69) — forming a chord — using the same Velocity value as before. Hold it for two beats, and then release.” When you play that back, the synthesizer follows the instructions and plays the audio part you wanted.

Why is MIDI so useful? For one thing, since MIDI files contain only instructions, the file sizes are a fraction of the size of digital audio files. Recording and playing them back is therefore relatively easy work, even with an older computer. In contrast, audio signals are very complex, so recording them digitally takes considerable computing power and results in a lot of data, so the files are quite large.

For example, depending on how many control instructions are included, a MIDI file of a stereo, 3-minute song might take up about 90 KB, and your computer can play it back with ease. But a stereo audio recording of that same song at CD quality (16-bit, 44.1 kHz) would be around 30 MB — more than 300 times bigger! And that's just with stereo recordings — when you get into multitrack recordings, the disparity between file sizes becomes enormous. If you had 20 instrument tracks of digital audio rather than just the left and right stereo tracks in that same 3-minute song, and you recorded on all 20 tracks throughout the entire song, the file size would be ten times larger, or 300 MB. Furthermore, an older computer might lack the horsepower to play all 20 tracks at once without glitches.

As a result, you can email and download MIDI files much more quickly than you can audio files, and you can save many more songs on your hard disk in MIDI format than you can in digital-audio format. You also can get by more easily with using an older computer if you use MIDI. On the other hand, digital audio has compelling advantages.

FIG. 2: The List view is a convenient way to see MIDI messages and values in detail.

Here's a more exciting reason to use MIDI tracks than merely saving disk space: MIDI allows students to instantly change the instrumentation. The MIDI Specification defines a Program Change message that tells a receiving device which of its internal programs to use. That allows you to tell a hardware or software synthesizer which sound to play. Although many synthesizers have proprietary sounds and program numbering, most modern digital synthesizers also support the General MIDI (GM) standard, which defines Program Change numbers for most of the common instruments and sound effects. For example, sending Program Change 1 to a synth's General MIDI sound bank always will cause the synth to play the sound of an acoustic grand piano. GM Program 9 gives you the sound of a celeste, GM 10 a glockenspiel, and GM 41 a violin. So you can change synth sounds simply by using a MIDI keyboard, software, or other controller to send the appropriate Program Change message to the synth.

For example, let's say you want to teach students to orchestrate a simple blues in the key of G, which will have a drum track and three other instruments. You have created the tune in a sequencer or notation program, and to get the students started, you have chosen a basic rock drum kit and finger-picked electric bass (GM 34), electric jazz guitar (GM 27), and alto sax (GM 66). Using Program Changes, the students could rapidly and easily change the sax to a trumpet (GM 57), the electric bass to an acoustic bass (GM 33), and the jazz guitar to an acoustic steel-string guitar (GM 26) — or to something more “out,” like a kalimba (101). They can also change the drums. (For more on MIDI in general and General MIDI in particular, see the feature story “Sequence-stration” in the Fall 2003 issue of MET, available at www.metmagazine.com.)

In contrast, with digital audio, an alto sax is always an alto sax, and although you can use an effects processor or EQ to alter the timbre (even twist it beyond recognition), you can't just push a button and make a sax into an oboe, as you can with MIDI.

Some things can be done with both MIDI and audio but are easier or better done with MIDI. For instance, let's say we are adding a vocal part to our simple blues. You are going to ask several students to take turns singing the part. The vocal parts will have to be recorded as digital audio, because we are capturing real acoustic singing, as opposed to recording instructions for a synthesizer. The backing instruments, however, could be MIDI tracks to be played back with a synthesizer or audio recordings of acoustic (or electroacoustic) instruments, or some combination of MIDI and audio. Let's see why MIDI might be the best choice here.

The key of G works fine for the first singer. The second student can't hit the low notes in G, but she could hit them if the song were transposed up to A. We could transpose audio tracks up a key but, depending on our software, that might change the timbre of the instruments in an unrealistic way. More advanced software lets you avoid those problems and change pitch without altering timbre or timing, but doing that for all of your background tracks can require some experimenting. Transposition is no problem with MIDI though, because the synth plays the same melody two semitones higher. You can transpose the whole song at once, with a single MIDI message. In the above example, the software will trigger an A instead of a G on the synth, but everything else will stay the same.

In most sequencing programs, you can edit individual notes or groups of notes in either MIDI tracks or audio tracks. Since a MIDI performance consists of individual MIDI events, however, students can easily change, edit, or delete a single note or a selected region of notes. With digital audio, they would have to edit single notes out of a continuous sonic stream which, although it can be done, is trickier.

In the music-technology lab, electronic keyboards are commonly used for MIDI input with notation and sequencing software. If available, consider using other types of MIDI instruments as well, such as MIDI guitar, wind, and percussion controllers triggering a sound module or software synthesizer. For example, if students play an electronic drum set to record the drum part for a MIDI sequence, the drum part will probably sound more realistic than if it had been entered from an electronic keyboard.

Whenever you want to record the actual sound of a musical performance (a complete piece or an individual instrument or voice), you make an audio recording. Some common reasons for using digital audio include recording rehearsals, concerts and audition CDs; documenting performance progress; and producing CDs to sell as fund-raisers.

We noted that your students' vocal parts on our example blues tune would have to be recorded as digital audio, because there was no practical way to add the vocal part with a synthesizer. (There are voice synthesizers, but they are difficult to use and inappropriate for this purpose.) Besides, the experience of singing the part is part of the lesson. The same will be true for your instrumental-performance students.

There are other practical reasons to choose digital audio over MIDI: even the best synthesizers are incapable of producing the detailed nuances that a live performer can elicit from an acoustic instrument. If your students want the sax part in a blues tune to sound like the real deal, complete with squeals and other saxophone effects, their best bet is to record a live sax player's performance as digital audio. The resulting track might not be as easy to edit, but it will sound more realistic. Students should record the vocals first, since the vocalist's range can decide the key of the song, and then record the sax player. That way, they won't have to transpose an audio track because of a key change.

In addition, a lot of hip-hop and other popular electronic dance music relies on sound clips that have been digitally recorded (sampled) from existing recordings; processed; looped, so that they will repeat continuously; and then combined with other sound clips to form entirely new pieces. Because the original sources of the sound clips are audio recordings, those sampled loops must also be in the form of digital audio. So if your students want to compose these types of musical pieces, they will need to work with digital audio.

Be sure, however, that the use of particular sound loops does not violate copyright laws. Some companies sell prerecorded audio loops for which all rights have been cleared, and they can be used in the classroom without worry.

Modern music technology provides teachers and students alike with a wide range of products and applications for using digital audio and MIDI. MIDI is neither superior nor inferior to digital audio; they are different tools. Once you understand what each approach is best for, you can choose the appropriate tool for each job.

Dennis Mauricio is the director of the Music Technology program at Hilltop High School in Chula Vista, California, and is the coauthor of Fundamentals of Music Technology (Consultant Help Software 3rd, rev. edition, 1999). Steve Oppenheimer is the editor in chief of MET and Electronic Musician magazines.