MIDI vs Audio: Instructions and Recordings

The single most useful distinction in a DAW: MIDI is instructions, audio is sound. A MIDI clip says "play C3, this hard, for this long". An audio clip is the sound — a recorded waveform. They live on different track types, edit differently and react differently to change.

What MIDI actually contains

MIDI (Musical Instrument Digital Interface) dates from 1983 and is still the standard. A MIDI note carries:

• Pitch — a note number from 0 to 127. Middle C is note 60 (named C3 or C4 depending on the DAW).
• Velocity — how hard it was played, 1 to 127. Instruments usually map it to volume and brightness.
• Start and length — in beats, so the part follows the project tempo.

Plus controller data: mod wheel, sustain pedal, pitch bend and so on. There is no sound in a MIDI file. Send the same clip to a piano and then to a synth bass and you get two completely different results.

What audio contains

An audio file is a sequence of samples — amplitude measurements taken thousands of times per second. Two numbers describe its resolution:

Setting	Common values	What it governs
Sample rate	44.1 kHz, 48 kHz	Highest frequency captured (about half the rate)
Bit depth	16, 24, 32-bit float	Dynamic range and noise floor

44.1 kHz captures up to about 22 kHz — beyond human hearing. 24-bit gives far more dynamic range than any room or microphone, which is why it's the standard for recording.

Side by side

	MIDI	Audio
Change a wrong note	Drag it	Pitch-correct or re-record
Change the instrument	Swap the plugin	Not possible
Change tempo	Free, follows the grid	Needs time-stretching, can add artifacts
Change key	Transpose, no quality loss	Pitch-shift, quality cost grows with distance
Capture a real voice or guitar	No	Yes
CPU use	Instrument plugins can be heavy	Playback is cheap

When to use which

Use MIDI for anything you're still deciding about: drum patterns, chord parts, basslines, melodies you might re-voice. It keeps every decision reversible.

Use audio for performances that only exist as sound — vocals, live guitar, a sample from a record — and for parts you've finished deciding on.

Committing: MIDI → audio

Most producers eventually bounce (render, freeze, print) instrument tracks to audio. Reasons:

• It frees CPU from heavy synths.
• It lets you chop, reverse and resample the result — classic sound design moves.
• It commits a decision, which helps you move on instead of tweaking forever.

Keep the MIDI track muted or archived so you can go back.

Extracting: audio → MIDI

The reverse is harder. Audio-to-MIDI conversion analyses a recording and estimates notes. It works well on clear monophonic material — a sung melody, a bass line, a lead synth — and less reliably on dense chords or full mixes, where overlapping harmonics confuse note detection. Expect to clean up the result: delete stray short notes, fix octave errors, and re-quantize.

Typical uses:

• Hum a melody into your phone, convert it, and play it back on any synth.
• Lift a bass line from a reference so you can study its rhythm.
• Double a recorded riff with a synth layer that follows it exactly.

If you want to separate one part of a full song first, isolate it with stem separation (for example the vocal remover for the voice) and convert the isolated part — the cleaner the input, the cleaner the MIDI.