Consider how MIDI data differs from digital audio as described in Chapter 5. You aren’t recording something through a microphone. MIDI keyboards don’t necessarily make a sound when you strike a key on a piano keyboard-like controller, and they don’t function as stand-alone instruments. There’s no sampling and quantization going on at all. Instead, the controller is engineered to know that when a certain key is played, a symbolic message should be sent detailing the performance information.
[aside]Many systems interpret a Note On message with velocity 0 as “note off” and use this as an alternative to the Note Off message.[/aside]
In the case of a key press, the MIDI message would convey the occurrence of a Note On, followed by the note pitch played (e.g., middle C) and the velocity with which it was struck. The MIDI message generated by this action is only three bytes long. It’s essentially just three numbers, as shown in Figure 6.7. The first byte is a value between 144 and 159. The second and third bytes are values between 0 and 127. The fact that the first value is between 144 and 159 is what makes it identifiable by the receiver as a Note On message, and the fact that it is a Note On message is what identifies the next two bytes as the specific note and velocity. A Note Off message is handled similarly, with the first byte identifying the message as Note Off and the second and third giving note and velocity information (which can be used to control note decay).
Let’s say that the key is pressed and a second later it is released. Thus, the playing of a note for one second requires six bytes. (We can set aside the issue of how the time between the notes is stored symbolically, since it’s handled at a lower level of abstraction.) How does this compare to the number of bytes required for digital audio? One second of 16-bit mono audio at a sampling rate of 44,100 Hz requires 44,100 samples/s * 2 bytes/sample = 88,200 bytes/s. Clearly, MIDI can provide a more concise encoding of sound than digital audio.
MIDI differs from digital audio in other significant ways as well. A digital audio recording of sound tries to capture the sound exactly as it occurs by sampling the sound pressure amplitude over time. A MIDI file, on the other hand, records only symbolic messages. These messages make no sound unless they are interpreted to do so by a synthesizer. When we speak of a MIDI “recording,” we mean it only in the sense that MIDI data has been captured and stored – not in the sense that sound has actually been recorded. While MIDI messages are most frequently interpreted and synthesized into musical sounds, they can be interpreted in other ways (as we’ll illustrate in Section 6.1.8.5.3). The messages mean only what the receiver interprets them to mean.
With this caveat in mind, we’ll continue from here under the assumption that you’re using MIDI primarily for music production since this is MIDI’s most common application. When you “record” MIDI music via a keyboard controller, you’re saving information about what notes to play, how hard or fast to play them, how to modulate them with pitch bend or with a sustain pedal, and what instrument the notes should sound like upon playback. If you already know how to read music and play the piano, you can enjoy the direct relationship between your input device – which is essentially a piano keyboard – and the way it saves your performance – the notes, the timing, even the way you strike the keys if you have a velocity-sensitive controller. Many MIDI sequencers have multiple ways of viewing your file, including a track view, a piano roll view, an event list view, and even a staff view – which shows the notes that you played in standard music notation. These are shown in Figure 6.8 through Figure 6.11.
[aside]The word sample has different meanings in digital audio and MIDI. In MIDI, a sample is a small sound file representing a single instance of sound made by some instrument, like a note played on a flute.[/aside]
MIDI and digital audio are simply two different ways of recording and editing sound – with an actual real-time recording or with a symbolic notation. They serve different purposes. You can actually work with both digital audio and MIDI in the same context, if both are supported by the software. Let’s look more closely now at how this all happens.Another significant difference between digital audio and MIDI is the way in which you edit them. You can edit uncompressed digital audio down to the sample level, changing the values of individual samples if you like. You can requantize or resample the values, or process them with mathematical operations. But always they are values representing changing air pressure amplitude over time. With MIDI, you have no access to individual samples, because that’s not what MIDI files contain. Instead, MIDI files contain symbolic representations of notes, key signatures, durations of notes, tempo, instruments, and so forth, making it possible for you to edit these features with a simple menu selection or an editing tool. For example, if you play a piece of music and hit a few extra notes, you can get rid of them later with an eraser tool. If your timing is a little off, you can move notes over or shorten them with the selection tool in the piano roll view. If you change your mind about the instrument sound you want or the key you’d like the piece played in, you can change these with a click of the mouse. Because the sound has not actually been synthesized yet, it’s possible to edit its properties at this high level of abstraction.
1 comment
Comments are closed.