Session Prep, Gain Staging & Mix Workflow: Fifteen Minutes That Decide the Next Ten Hours

Open a session someone sends you. Forty-one tracks named "Audio 07.3" and "New Recording 12", half of them starting at different points, one at the wrong sample rate, a vocal peaking at 0 dBFS and a bass so quiet you wonder if it was recorded. You could start turning knobs. Every experienced engineer does something else first: they spend the first fifteen minutes making the session a place they can think in.

That quarter hour decides how the next ten hours go. Mixing is the craft of decisions about sound, and decisions go wrong when the session fights you: when you cannot find the tom tracks, when a plug-in distorts because the level feeding it is wrong, when you have been listening for three hours at a volume that is tiring your ears and flattering the bass. This piece is about everything that happens before the "creative" mixing, and about the order in which a mix gets built. It is the first of the practical pieces of the Mixing & Mastering specialization, so it ends with a checklist you can use at the desk.

Receiving the session: check before you touch anything

Files arrive from a tracking session, a band's own laptop, a producer in another country or your own earlier project. The checks are the same. First, the format: the sample rate and bit depth the files were recorded at (Digital Audio & the DAW, 36/62). Mix at the rate the files already have, usually 44.1 or 48 kHz at 24-bit. Converting is a decision to make deliberately, not by importing blindly.

Importing at the wrong rate is the classic accident. If a 44.1 kHz file is played as though it were 48 kHz, every sample comes out faster: the song shortens and the pitch rises by 12 × log₂(48 ÷ 44.1), about 1.46 semitones. The reverse slows and lowers it. Most DAWs now convert or warn, but exported "stems" from other programs can still arrive with a mismatch. The tool below shows the size of the error.

Second, start points and lengths. Every track exported from a session should start at the same moment (bar 1, or time zero) so they line up the instant you drop them in. A track that was trimmed and exported from where the first note begins will sit at the wrong place, and nobody will hear it until the vocal arrives a bar early. Third, completeness: is every instrument the band remembers actually present, with takes that are the right ones? Ask for the tempo, a rough mix and a note of anything special (a deliberate distortion, a sound that should be that harsh) before you spend an hour "fixing" it.

Naming, colour and buses: a session someone else could open

Rename every track to what it is: Kick In, Kick Out, Snare Top, Snare Bottom, OH L, OH R, Bass DI, Bass Amp, Gtr Rhythm L, Lead Vox. Group them by family and give each family a colour, in the order you will meet them: drums, then bass, guitars, keys and synths, vocals, effects. A drummer's seven tracks should look like one object. Delete or archive takes you will never use, but never delete the originals: keep the delivered files in their own folder and work on copies. That is also your backup.

Then build the bus structure. A bus sums several tracks so you can treat them together. A typical rock layout has a drum bus (all drum tracks, perhaps with a sub-bus for the kick and snare), a bass bus, a guitar bus, a keys bus, a vocal bus (lead on one track, doubles and harmonies on a sub-bus) and one or two effect returns, a reverb and a delay that tracks send into. Everything ends at the mix bus, sometimes called the master or stereo bus. The point of the structure is control: a single fader moves all the guitars against the voice, and one compressor on the drum bus glues them. Pop and hip-hop sessions often have fewer instrument buses and many more vocal tracks; the shape of the tree follows the music.

DAWs differ on the vocabulary (REAPER has folders and tracks that act as buses, Pro Tools has aux inputs and VCA masters, Logic has summing stacks and aux channels, Ableton Live has group tracks and return tracks, FL Studio has mixer inserts routed to other inserts), but they all let you do the same four things: group, route, name and colour.

Gain staging: where the numbers live

Gain staging means setting the level of the signal at each point in the chain so nothing is too quiet or too loud for what comes next. In the analogue world each device had a noise floor below and a clipping point above, and you kept the signal between them. A digital file has the same two limits at the output end: the quietest bit the file can store, and 0 dBFS, the loudest sample it can hold. But the chain inside a modern DAW is different.

Most DAWs add and process tracks in 32- or 64-bit floating point. A 32-bit float number has a 24-bit mantissa and an 8-bit exponent, so it can represent values far above 0 dBFS: about 1,500 dB of range in total, against the roughly 144 dB (6.02 × 24) of a 24-bit integer file. So a channel inside the DAW that sits 10 dB "over" is not clipping. Nothing is lost, and turning it down restores it exactly. Two places do clip: a plug-in that expects a certain input level (many clip or behave nonlinearly beyond it), and the output converter or the exported file, which are fixed-point. The rule is therefore not "never exceed 0 dBFS inside the mix bus" but "keep every plug-in input sensible and keep the final output under 0".

The number that comes up is −18 dBFS. In a digital system the standard alignment level is −18 dBFS (EBU Recommendation R68, which the digital broadcast world adopted), the level that corresponds to a nominal +4 dBu, or 0 VU, on professional analogue gear, leaving about 18 dB above it before digital full scale. Plug-ins that model analogue compressors and tape are generally calibrated so that their "0 VU" sits at a level around there, and they react to level the way the hardware did, saturating harder when you push them. Sound On Sound's gain-staging article advises setting the loudest source so that it peaks between −12 and −18 dBFS, and says modelled plug-ins "start to fall down" when pushed far outside their normal operating range. Check your plug-in's manual: calibration differs between makers, and the −18 figure is a convention, not a law. Your ears and the plug-in's own meters overrule it.

That arithmetic is why hot tracks matter. If the kick, snare, bass, guitar and voice are each recorded near full scale, the bus climbs several decibels above 0 before you have moved a fader. Try the five synthesised tracks below. They are generated in the page, each recorded at its own level, and summed the way a bus sums them. Watch the bus peak, the number of samples that would clip at the converter, and which tracks sit in the −18 to −12 dBFS zone.

On the "as delivered" preset the bus peaks above 0 and the converter clips: you may hear an edge on the kick and bass where they land together. Pull the master fader back 6 dB and it is clean again, which shows the point about floating point: the information was never lost, only the output was too hot. The better fix is at the source: trim each track so the loudest one peaks between −12 and −18 dBFS, then use the faders to balance. Do it with the track's trim or clip gain, not a plug-in, so every plug-in downstream sees a sensible level.

Listen first, then balance, then process

Before any processing, play the song from the top, once, with every fader at its starting level. Listen to the arrangement: where it builds, what carries the interest in each section, which part is the hook. Bobby Owsinski's The Mixing Engineer's Handbook describes six elements of a mix (balance, frequency range, panorama, dimension, dynamics and interest), and the order is not accidental: balance comes first because nothing else works without it.

If you were given a rough mix, play it. It tells you what the band and producer heard and want, and it is a reference for the arrangement. You are not obliged to copy it, but ignoring it is a choice you should make deliberately. Then build a static balance with faders and pan alone. Zero every plug-in (or bypass them), set the tracks to roughly the right level against each other, and pan. Many engineers do this in mono first: a balance that works in mono has fewer masking problems to unmask later. Only then reach for EQ and compression. The reason is practical. A track that sounds thin on its own in solo may be perfect once the others are in; the purpose of processing is to fix problems that stop the balance working, and you cannot see the problems until the balance exists. Solo is a tool for finding clicks and noise, not for judging tone.

Save a version before you start and after each stage. Name them plainly: "Song_mix_v01 static balance", "v02 drums done". A mix that goes wrong at hour six should cost you ten minutes, not the afternoon.

Monitoring level, references and rest

What you hear depends on how loud you listen. Human hearing is less sensitive to low and high frequencies at low levels, a result first measured by Fletcher and Munson in 1933 and now standardised as ISO 226. Mix quietly and the bass and treble sound weak, so you boost them; the track then sounds boomy and harsh when it is played louder. Mix too loud and the ears tire within the hour, which makes everything sound duller and encourages brightening. The accepted compromise is a fixed, moderate monitoring level that you can return to.

There is a published way to define it. Bob Katz's K-system calibrates each loudspeaker with pink noise at −20 dBFS RMS to 83 dB SPL (C-weighted, slow response), the level he proposed for wide-dynamic-range work; he notes that his typical pop monitor gain was 6 dB below that, which yields 77 dB SPL with the same noise. Film mixing rooms are calibrated by a similar method, with a figure often quoted near 85 dB SPL per channel. They are measured in large, treated rooms. A small bedroom studio has strong bass build-up, so many engineers there listen lower, and a quiet balance in a small room is a very different thing from a loud one. Treat 77 to 85 as the reference range for a calibrated room, not a target for your bedroom. Use an SPL meter app as a rough guide, then settle on one setting and mark the knob.

Loud listening is also a hearing-safety question. The World Health Organization's safe-listening guidance takes 80 dB(A) for about 40 hours a week as the adult reference, and every 3 dB more halves the safe time. This is general information, not medical advice. Do your loud checks briefly.

The second habit is the reference track: a finished record you know well, in the genre you are mixing, played in your room at the same loudness as your mix. If you switch to a louder reference, it will sound better for that reason alone, because our ears favour louder, and your mix will seem dull, so you will brighten it for no good reason. Match the level first, then compare the tone, the low end, the width and the vocal level. Mastering & Loudness (36/64) covers LUFS; here a simple average level is enough to show the trap. The tool below builds a "reference" from your own mix by making it more dense, so it is louder at the same peak. Switch matching on and off.

Finally, rest. Ears and attention both fade. Many engineers take a short break every hour or so, and end a long day by saving, then leaving the mix until tomorrow. A listen after a night away finds things that hours of tweaking could not. Never judge a mix on the last thing you did to it.

The order a mix is built in

There is no single correct order, and engineers argue about it. What matters is that you have an order, so you do not wander. The most common starting point is a balance-first order: faders and pan, cleanup, drums and bass, then harmony instruments, then the vocal, then effects, then automation and the bus. Three alternatives are widely taught. Drums-first builds the kit and bass until the foundation is powerful, then adds the rest in order of importance. It suits rock, where the rhythm section carries the record. Vocal-first sets the singer's level and tone against a rough backing, then builds everything else to support the voice. It suits pop and rap, where the voice is the product. Top-down starts with the stereo-bus processing already on and mixes into it, so you hear the mix as it will be, and adjusts that processing as you go. Its critics say it hides problems; its fans say it avoids surprises. Try them all on the same session: the one that feels clearest to you is the one to keep.

Starting points, not rules

These are typical figures from the sources named. They are where people begin, not where they must end up. Your ears, the song and the room overrule every one.

The reason your ears overrule them: a range describes typical material through typical gear in a typical room. The tracks in front of you are not typical, and the figures only save you time on the first guess.

Common mistakes

Genre and world differences

The routine above is the same everywhere; the content of the session changes. A rock or metal session arrives with a multi-miked drum kit, two or four guitar tracks per part and a lot of microphone bleed; drums-first is natural and editing the drums to tightness is a first job. A pop session often has a programmed beat, dozens of vocal tracks and vocal-first order. In hip-hop and trap the producer usually sends a stereo beat or a few stems with the vocals recorded on top, so the mix is vocal against beat, and the 808's low end is the first level decision. Reggaetón and dembow productions work the same way, with the dembow pattern and its bass balanced against a vocal that sits very close to the front, and many tracks of ad-libs and doubles. In cumbia the layers of percussion, bass and melodic instruments (accordion, guitars or synthesisers depending on the style) are dense in the midrange, so the static balance and the pan do most of the work. Jazz and acoustic folk sessions are small: the question is the balance of a few live instruments and the room, and the correct amount of processing is often very little.

Afrobeats and amapiano, like much modern pop worldwide, often come from producers' own sessions of programmed drums and layered percussion, so the engineer receives stems and works vocal-first; highlife and other live-band traditions arrive as many-microphone sessions that look like rock sessions with different instruments.

The session-prep checklist

In the order you do them at the desk. Tick as you go; copy it as text for your notes; print it for the wall.

Practise on a real session

The best way to learn this routine is to run it on a real multitrack. Mike Senior's Mixing Secrets website, Cambridge-MT, hosts a free multitrack download library of several hundred projects. He describes the files as free to download and use for mixing practice; they are 24-bit, 44.1 kHz WAV files in ZIP archives, each with an MP3 preview, and the library has a search page by style. Read the terms on the site before you post a mix publicly or use one for anything beyond practice. Pick a style you listen to, and for a first attempt choose a session with a modest number of tracks. Run the session-prep checklist, rename and colour the tracks, build the buses, balance with faders, and only then touch a plug-in.

What a graduate just sees

• The first fifteen minutes (names, colours, buses, levels) decide the next ten hours.
• A float-point DAW does not clip inside, but plug-ins and the output converter do; set levels for them.
• −18 dBFS is the digital alignment level that analogue-modelled plug-ins are built around; peaks of −12 to −18 on import are the usual advice.
• Balance with faders and pan before any processing; solo tells you about noise, not tone.
• You choose a monitoring level and keep it; small rooms and long sessions both argue for quieter.
• A louder reference sounds better for that reason alone, so match levels before comparing.
• There are several workable build orders; having one matters more than which.
• Save versions, never touch the originals, and rest your ears.

Where it shows up next

• Mixing Drums (36/85): the drum bus, the tracks named in step one, and the phase check that comes right after balance.
• Mixing Bass & Guitars (36/86) and Mixing Vocals (36/87): the same session, one family at a time.
• Finishing the Mix (36/88): the translation checks and the delivery headroom that these gain-staging habits make easy.
• Mastering & Loudness (36/64): the reference-level comparison reappears with true loudness measurement.
• In the world: any podcast, voiceover or video edit has the same session, level and monitoring habits.

If you want to go deeper

• 📘 Textbook. Mike Senior, Mixing Secrets for the Small Studio (Routledge, 2nd edition, 2018), a practical course for mixing in a home studio; and Roey Izhaki, Mixing Audio (Routledge), a thorough treatment of every tool.
• 📘 Short and practical. Bobby Owsinski, The Mixing Engineer's Handbook (Cengage), on workflow and the elements of a mix.
• 🌐 Free practice. The Cambridge-MT free multitrack library, to run the checklist on a real session.
• 🌐 Free reading. Sound On Sound's gain staging in your DAW article; Bob Katz's Level Practices on K-system monitoring.