Site last updated: Dec-2002
All material on this site is copyright © The Author

Music may be stored in one of two ways. An audio recording or an electronic representation of the original music.

In an audio recording, the music is usually transferred to the recording medium by placing a microphone in front of the musicians, or by a direct connection to an acoustic pick-up mounted on the instrument. The sound is converted to an electrical signal, amplified and recorded on a tape or a digital recording medium, for example CD. Now, most people understand the basics of this process. Take a tape recorder, connect a microphone, aim it at the instruments and, well, that’s it! But when it comes to music stored on a computer as, for example MIDI, then their understanding of the process is either non existent, or at least, vague, simply because they can not visualize what is going on.

MIDI (Musical Instrument Digital Interface) does not store music as sounds. It uses a system of numbers, which represent sounds, when they are fed to a device which ‘understands’ what the numbers mean. This ‘device’ is called a ‘Synthesizer’, or ‘Synth’ as they are commonly known.
There are two basic types, ‘Digital’ and ‘Analogue’. The‘Analogue’ variety operate with varying voltages. MIDI can operate these devices, through the use of a special control unit which converts the digital MIDI information to analogue voltages. The digital type can process MIDI directly and is in more common usage.

Most PC’s have a sound card fitted, which contains a digital synthesizer. When the synth is fed with MIDI, it uses digital signalling techniques to translate the MIDI into musical instrument sounds.

MIDI may also be used to play mechanical music instruments. A microprocessor circuit can be designed to read MIDI and operate solenoids, or servos, which are coupled to the various components of a mechanical instrument. See the section on Solenoids for some idea’s on the subject.

Q: So what is the difference between using audio and MIDI to store music?

A: MIDI has an advantage over audio, in that it occupies considerably less space for a particular length of ‘Song’. If you were to take a typical MIDI file of, say, 25000 bytes (25K bytes), play it through a synth and record the resulting music on an audio system, it would probably take up at least one hundred times as much storage space-and that would depend on the quality required! Even with compression, for example MP3, the audio file would be substantially larger than the MIDI file.

The main disadvantage with MIDI is that it only gives an approximation of the composer’s wishes. In acoustic music, the recording is exactly the same as the original performance, after some tweaking on the mixing desk. With MIDI, the sounds produced, depend on the quality of the synthesizer.

As technology progressed, more ‘natural’ sounding synthesizers became available. The ‘Sampler’ became the standard. In this system, real sounds are recorded as digital audio and played back when MIDI is fed in. This means that the sound of a real organ pipe could be stored in the sampler and played back with no loss of quality or ‘Timbre’. The sound is not synthesized, it is literally being played back from the digital audio. The pitch of each sample may be altered, so that one organ pipe could, in theory, be made to play back ‘Polyphonically’ over the full spectrum. The problem with this is that a natural sound only sounds natural at the pitch it was sampled. So in an ideal system, each note of a scale should be sampled separately (usually only one or two notes are assigned to each sample).

Q: We have heard how MIDI is turned into audio, but is it possible to turn audio into MIDI?

A: There have been numerous attempts at providing this, but none of them have been very successful. This is a very technical problem. MIDI is stored as a set of numbers, which makes it relatively easy to recreate the music it represents. Audio, on the other hand, is a complex set of sound waves. Also, remember that most musical instruments produce ‘Harmonics’ (multiples of the fundamental frequency/pitch), which gives each instrument its characteristic sound or timbre. The only way to interpret audio is to carry out a very complex frequency analysis and even then, it would require that the analyzer ‘knew’ how to interpret each type of instrument in the recording. It would have to know which harmonic belonged to which instrument.

I am sure that as technology moves forward, we will eventually see a relatively cost effective audio to MIDI converter-but don’t hold your breath! Of course, if you know otherwise, please let me know-maybe there already is one ’out there’!