GlossaryBeginnerAudio

Stems, Mixing, and Mastering: A Beginner's Glossary

Plain-English definitions of the audio terms you keep seeing — stems, dry/wet, headroom, sidechain, LUFS — explained without jargon and with real-world examples.

ONYX Engineering////12_MIN_READ

If you are new to making music, half the words people throw around sound like a different language. Here is what the most common ones actually mean, with examples you can hear when you listen to your favorite tracks.

This is a living glossary. We add to it. Bookmark and come back.

Stems

A stem is one isolated layer of a song — just the vocals, just the drums, just the bass. A finished song is usually a mix of 4–8 stems. The standard 4-stem split is:

  • Vocals
  • Drums
  • Bass
  • Other (everything else: keys, synths, guitars, FX)

Stems matter because they let you re-mix or remaster a track later without re-recording it. They are also required for sync placements (so a music supervisor can duck the vocal under dialogue) and remix releases.

Most AI tools only export the final stereo file. Some (including ONYX Pro on every render) export stems separately as 44.1 kHz / 24-bit WAV files.

Mixing vs Mastering

These get confused constantly. They are different jobs.

  • Mixing is balancing the individual elements of a song against each other. Vocal sits above the snare, bass holds the bottom, hi-hats sparkle on top. The mix engineer is making the song work as a coherent piece of music.
  • Mastering is the final polish on the already-mixed song. The master engineer is making it work everywhere — in your headphones, in a car, on a phone speaker, on a club system, after Spotify's encoder squeezes it through Ogg Vorbis.

You mix to make the song right. You master to make it travel. (For the specifics, see our Spotify mastering guide.)

Dry / Wet

The dry signal is the unprocessed sound — your vocal with nothing on it. The wet signal is the processed version — vocal plus reverb, plus delay, plus whatever else. A "50% wet" reverb means you hear half dry vocal, half reverb tail.

Most modern vocals run something like 80% dry, 20% wet for the lead and 60/40 for the harmony stack. Going past 30% wet on a lead usually pushes the vocal too far back.

Headroom

The space between your loudest peak and digital zero (0 dBFS). If your kick peaks at -3 dBFS, you have 3 dB of headroom. Mastering engineers usually want -6 dB of headroom in the mix they receive, so they have room to add their own processing without clipping.

If you deliver a mix at -0.1 dBFS peak, the mastering engineer has no room to work and has to ask for a quieter version. Always leave headroom.

Sidechain

A compressor that ducks one sound out of the way of another. Classic example: the kick drum sidechains the bass, so every time the kick hits, the bass drops 2–3 dB. The kick punches through, the bass fills the space when the kick is gone.

House, dance, and modern pop live on this trick. The "pumping" feel of EDM is almost always a sidechain compressor pulling 4–6 dB off the pads every time the kick fires.

LUFS

Loudness Units relative to Full Scale. It is the modern standard for measuring perceived loudness — much more accurate than the old peak meters.

  • Spotify normalizes to about -14 LUFS integrated
  • Apple Music to about -16 LUFS
  • YouTube to about -14 LUFS
  • TikTok / Reels run hotter, around -9 LUFS

If your master is louder than the platform's target, the platform turns it down. There is no benefit to mastering hotter than the platform plays. See the Spotify mastering guide for the practical implications.

True peak (dBTP)

The loudest inter-sample peak after digital-to-analog conversion. Different from sample peak. Limiter ceilings should be set to -1 dBTP, not -0.1 dBFS, to absorb codec inflation.

Transient

The very beginning of a sound — the click of a kick, the attack of a snare, the pluck of a guitar pick. Transients are what give a mix punch. Compress them too hard and the mix sounds flat; let them breathe and the mix sounds alive.

A transient shaper is a plug-in that lets you boost or cut transients without touching the body of the sound. Useful on drums.

EQ (Equalization)

Boosting or cutting specific frequency ranges. Common moves:

  • High-pass filter — removes everything below a chosen frequency. Used to clean mud out of vocals around 80–120 Hz.
  • Low-pass filter — removes everything above a chosen frequency. Used to take air off a sound that is fighting the lead vocal.
  • Shelf — boosts or cuts everything above (or below) a point.
  • Bell — targets a narrow band around a center frequency.
  • Dynamic EQ — only engages when the input crosses a threshold (e.g. tame harshness only when it appears).

The first rule of EQ: cut before you boost. A 2 dB cut almost always sounds more natural than a 2 dB boost.

Compression

Reduces the dynamic range of a sound. The four key controls:

  • Threshold — the level above which the compressor engages.
  • Ratio — how much it reduces signal above the threshold (4:1 means 4 dB above threshold becomes 1 dB).
  • Attack — how quickly the compressor responds.
  • Release — how quickly it lets go.

Slow attack preserves transients. Fast attack squashes them. Use accordingly.

Reverb / Delay

  • Reverb simulates a space — a small room, a hall, a cathedral. It adds depth.
  • Delay is a discrete echo — the sound repeats one or more times. It adds rhythm.

Most modern vocals use a touch of both: short reverb for body, slap delay (one quick echo around 80–120 ms) for character.

Saturation

Subtle harmonic distortion added to a sound to make it feel warmer, fuller, or more present. Tape saturation, tube saturation, and transformer saturation are the three classic flavors. A touch on the master bus adds perceived loudness without crushing transients.

BPM

Beats per minute. The tempo of the song. 120 BPM means two beats every second. See our BPM landing pages for genre-specific tempo guides.

Key

The tonal center of a song — the note that feels like "home." Songs in C Major feel different from songs in C Minor. Most pop sits in five or six common keys. Our key landing pages walk through the mood of each.

Time signature

How many beats are in a measure and what counts as a beat. 4/4 means four quarter-note beats per bar — the most common time signature in modern music. 6/8 has a triplet feel. 3/4 is waltz time.

Sub bass

The lowest frequency range, roughly 20–60 Hz. You feel it more than you hear it on small speakers. In trap and drill, the 808 lives here. In dance music, the kick lives here.

Stem mastering

A halfway point between mixing and mastering — the master engineer receives stems instead of a single stereo bounce. This lets them shape individual elements rather than just the whole. More expensive than regular mastering, but useful when the mix needs finishing touches the engineer can address per-stem.

Loudness war

The decade-long trend (roughly 1995–2015) where every commercial release got mastered louder than the one before, until masters became fatiguing brick walls with no dynamic range. Streaming loudness normalization ended the war, but its scars are still in your record collection.

More to come

We add definitions as we hear new ones come up in support tickets and Discord. If a term is not in here, ping us and we will add it. The goal is to make the entire vocabulary of music production legible to someone who started yesterday.

// GENERATE WHAT YOU JUST READ ABOUT