How to Prepare Stems for Online Mixing

A practical stem export checklist before sending your track for online mixing.

Good stem exports make online mixing faster, cleaner and less frustrating for everyone involved.

The goal is simple: every file should line up from the same start point and rebuild the rough mix when imported into a new session. If the engineer drops your files into a DAW and the song immediately plays in the right place, you have done half the boring work correctly.

And boring work matters.

Bad exports waste time. Missing files slow the process down. Random file names create confusion. Stems starting from different places can turn a simple mix into a detective story nobody asked for.

Before sending your track for online mixing, take a little time to prepare the session properly. It helps the engineer understand your song, protect the parts you already like and focus on improving the record instead of cleaning up chaos.

Stems, multitracks and what people usually mean

First, a small language issue.

Technically, stems are grouped audio files. For example:

• drums
• bass
• vocals
• guitars
• synths
• effects

Multitracks are individual tracks. For example:

• kick
• snare
• hi-hat
• lead vocal
• backing vocal 1
• bass DI
• bass amp
• rhythm guitar left
• rhythm guitar right

In real life, many artists use the word “stems” for everything. That is fine. Nobody needs to be arrested by the audio police.

For full online mixing, individual tracks are usually better. They give the engineer more control. For stem mastering, grouped stems are enough. For full mixing, send the detailed files whenever possible.

If you are not sure, send a rough mix and ask before exporting the whole jungle.

Start with a clean session

Before exporting anything, clean the session.

Remove obvious muted tracks, unused takes, empty audio clips and random experiments that are not part of the final song. If you have 47 versions of the same synth idea and only one is actually used, do not send all 47 unless you enjoy watching engineers age in real time.

A clean session helps avoid mistakes.

Keep the tracks that matter. Remove the tracks that do not. If something is muted because it is not part of the song, leave it out. If something is muted but important as an option, label it clearly.

The engineer should not have to guess what belongs in the record.

Export every file from the same start point

This is the big one.

Every track should start from the same point, even if the instrument does not play until later in the song.

For example, if the vocal starts at 0:45, the exported vocal file should still begin at the same start point as the rest of the song. It should include silence before the vocal enters.

Why?

Because when all files are imported into a new session, they line up instantly.

If every file starts at a different moment, the engineer has to manually place them. That creates risk. One small timing mistake and suddenly the percussion feels lazy, the backing vocal is late, or the guitar sounds like it had a long lunch.

Export from bar one, song start, or the same timestamp for every file.

No exceptions unless agreed in advance.

Use proper audio format

Send WAV or AIFF files.

Preferably:

• 24-bit WAV
• same sample rate as the project
• no MP3
• no random online converter files
• no screen-recorded audio
• no files pulled from a messaging app

MP3 is not a mixing format. It is a delivery or preview format. Sending MP3 stems for mixing is like bringing soup in a paper bag. Technically brave, spiritually questionable.

If your project is 44.1 kHz, export 44.1 kHz. If it is 48 kHz, export 48 kHz. Do not upsample just to look fancy. Clean and consistent is better than fake impressive.

Keep important creative effects

Not every effect should be removed.

If an effect defines the sound, print it.

For example:

• guitar amp tone
• vocal tuning
• vocal distortion
• creative delay throws
• sound-design effects
• synth processing
• special filters
• risers and transitions
• sidechain pumping that is part of the groove

If the effect is part of the identity of the track, keep it.

The engineer is not there to accidentally erase your personality. If the distorted vocal is the whole vibe, do not send only a dry version and hope someone guesses the original intention.

A good rule:

If the effect is creative, print it.
If the effect is only rough mixing, ask or send both.

Send dry and wet versions when useful

Sometimes the best option is to send both processed and unprocessed versions.

This is especially useful for:

• lead vocals
• guitars
• important synths
• special effects
• tuned vocals
• AI-generated or AI-separated vocals

For vocals, a dry version can give the engineer more control. But if your rough vocal sound has a delay, distortion or special texture that matters, include a wet version too.

Label them clearly:

• Lead Vocal Dry
• Lead Vocal Tuned
• Lead Vocal FX
• Lead Vocal Wet

Do not call everything “audio 7”. That is not a file name. That is a cry for help.

Print tuning, timing and sound-design decisions

If tuning and timing are already part of the production, print them.

This includes:

• tuned vocals
• edited vocal timing
• corrected guitar timing
• MIDI instruments bounced to audio
• sound-design layers
• rhythmic effects
• printed sidechain movement
• special transitions

Do not assume the engineer will recreate your plugin chain unless that has been agreed.

If the sound is important, commit it.

There is an old studio truth here: options are useful until they become fog. A clear decision is often better than sending twelve half-finished possibilities and calling it “flexibility”.

Leave headroom, but do not overthink it

Do not export stems clipping into red.

Keep levels healthy and clean. You do not need to make every file tiny. You also do not need to normalize every track to maximum level.

A sensible export should have enough headroom and no obvious clipping.

Avoid:

• clipping individual tracks
• normalizing every stem to 0 dB
• exporting through a master limiter
• making files unnecessarily loud
• changing balances randomly before export

If the rough mix sounds right, keep the production balance sensible. The engineer can adjust levels during mixing.

The point is not to make the stems loud. The point is to make them clean.

Remove master bus processing unless agreed

This one depends on context.

For full mixing, do not export every track through heavy master bus processing unless it is part of the sound and has been discussed.

A limiter, clipper or heavy master chain can make the exported tracks behave strangely. It may also hide problems or make the engineer fight processing that should not be there.

However, if your master bus processing is a major part of the production vibe, send a rough mix with it. That helps the engineer understand the direction.

Good practice:

• send individual tracks without master limiting
• send a rough mix with your current mix bus sound
• mention any important master chain processing
• include references if you have them

That way the engineer knows what you liked, without being trapped inside your limiter settings.

Name files clearly

Clear file names save time.

Use simple names like:

• Kick
• Snare
• Hi Hat
• Bass
• Lead Vocal
• Backing Vocals
• Guitar L
• Guitar R
• Piano
• Synth Lead
• Pads
• FX
• Drum Loop
• Percussion

If there are multiple versions, label them properly:

• Lead Vocal Dry
• Lead Vocal Wet
• Lead Vocal Double
• Lead Vocal Harmony
• Bass DI
• Bass Amp

Avoid file names like:

• final stem
• new audio
• track 14
• export final final
• maybe use this
• vocal good one maybe
• really final fixed now

Your future self deserves better. So does the engineer.

Include a rough mix

Always include a rough mix.

The rough mix shows the intention. It tells the engineer what you have been hearing while building the track. Even if the rough mix is technically messy, it can be very useful.

It helps with:

• vocal level intention
• arrangement balance
• effects direction
• section energy
• emotional shape
• what parts matter most
• what should not be “fixed” away

Without a rough mix, the engineer may technically mix the files correctly but miss the mood.

A rough mix is the map. The stems are the road.

Send both.

Include reference tracks

References are useful, but they should be chosen carefully.

Send one to three tracks that show the direction you want. They can help explain tone, width, vocal level, low-end style or overall energy.

But do not send ten references that all contradict each other.

If you send one track for vocal sound, another for drums, another for low end, say that clearly.

For example:

• Reference 1: vocal level and intimacy
• Reference 2: drum punch
• Reference 3: overall darkness and width

This is much more useful than saying “make it sound expensive”. We all want expensive. The question is which kind of expensive.

Include tempo and key if possible

Tempo and key are not always required, but they help.

Send:

• BPM
• key
• time signature if unusual
• lyric sheet if vocals matter
• notes about arrangement changes
• any problem areas you already hear

This is especially useful when the track has tempo-based delays, editing work, vocal tuning, MIDI rebuilds or production changes.

If you know something is important, say it.

Do not make the engineer discover the secret bridge like it is hidden treasure.

What about AI-generated or AI-separated stems?

If the stems come from AI tools, say so.

That is not a shameful thing. It just changes the technical reality.

AI-generated or AI-separated stems can contain:

• bleed between instruments
• watery artifacts
• unstable vocals
• smeared drums
• fake stereo width
• odd phase issues
• missing transients
• damaged high end

The engineer needs to know what kind of source material they are working with.

If you used an AI stem splitter, include the original full mix as well. Sometimes the original file contains useful information that the separated stems damaged or lost.

For AI-assisted tracks, good preparation may include:

• original AI export
• separated stems
• rough mix
• any edited or rebuilt parts
• notes about what should be improved
• notes about what should stay close to the original

Honesty here saves time.

Nobody needs a dramatic confession. Just label the files properly and explain the source.

Do a quick import test before sending

Before sending the files, test them.

Create a new empty session and import all exported tracks. Press play.

Ask yourself:

• Do all files line up?
• Does the rough mix rebuild correctly?
• Is anything missing?
• Are any tracks silent?
• Are the vocals in the right place?
• Are the effects printed correctly?
• Are the files named clearly?
• Is the export clean from start to finish?

This simple test catches many problems before the engineer ever downloads the folder.

It takes a few minutes. It can save hours.

That is a good trade.

How to package the files

Put everything in one clean folder.

A useful structure:

• Audio Stems
• Rough Mix
• References
• Notes

Inside the notes, include:

• artist name
• song title
• BPM
• key
• sample rate
• any important effects
• what you like about the rough mix
• what you want improved
• any known problems

Then zip the folder before sending.

Use a proper file transfer link. Do not send 38 separate email attachments like it is 2007 and everyone is suffering together.

What not to send

Avoid sending:

• MP3 stems
• clipped exports
• random start points
• unlabeled files
• missing vocals
• missing effects
• mastered stereo file only, if you want full mixing
• files exported through heavy master limiting
• ten different versions with no explanation
• AI-separated stems without the original file

A mixing engineer can solve many things.

Mind-reading is not one of the services.

At least not yet. Probably for the best.

Final checklist before sending

Before sending stems for online mixing, make sure:

• all files start from the same point
• all files are WAV or AIFF
• files are preferably 24-bit
• sample rate matches the project
• no tracks are clipping
• important creative effects are printed
• dry and wet versions are included where useful
• tuning and timing decisions are printed
• files are named clearly
• rough mix is included
• references are included
• BPM and key are included if known
• the exported files rebuild the song in a new session

If this is done properly, the mixing process becomes much smoother.

Final advice

Good stem preparation is not glamorous, but it is professional.

It helps the engineer spend less time organizing broken exports and more time making the song sound better. It protects your creative intention, reduces confusion and gives the mix the best chance of starting in the right place.

At Unsaid Records, online mixing works best when the files are clean, clearly named and exported from the same start point. If the track comes from a traditional DAW session, AI-assisted workflow or separated stems, the goal is the same: make the material clear enough to mix properly.

No messy folder archaeology. No guessing games. Just a clean handover so the record can move forward.