What files do I need to send to a mixing engineer?

Three things. First, your vocal recording with no effects applied, exported as WAV. Second, the instrumental you recorded against, in the highest quality version you have. Third, a reference track or link showing the sound you are aiming for. Everything else, including harmonies and ad-libs, should be exported as separate files using the same start point as the main vocal.

What does 2mix mean in Japanese mixing terminology?

Literally it means audio that has been summed to two channels, in other words a finished stereo file. In practice Japanese engineers use it loosely and it can mean either a stereo instrumental or a rough mix of your vocal over the instrumental. Because the meaning shifts between people, if a Japanese engineer asks you for a 2mix it is always worth confirming exactly which file they mean before exporting.

Should I apply reverb or autotune before sending my vocals?

No. Send the recording exactly as it came out of your microphone. Reverb, compression, EQ and pitch correction cannot be removed once they are printed into the file, and every one of them limits what the engineer can do. If you like a particular effected sound, send a separate reference version alongside the clean take rather than sending only the processed file.

What export settings should I use for my vocal tracks?

Export as WAV at 24-bit and at the same sample rate as your project, normally 48 kHz. Do not export as MP3, because the compression permanently removes high frequency detail that pitch correction and de-essing rely on. Do not normalise or apply any loudness processing. If your peak level sits somewhere between about minus twelve and minus six decibels, that is comfortable.

Why does every track need to start at the same point?

Because that is what keeps everything in sync. Export every track from the very beginning of the project rather than from where the singing starts, so that all files are the same length and line up automatically when dropped into a session. If tracks are trimmed to their first note, the engineer has to align them manually by ear, which takes time and introduces the risk of timing errors.

How does sending a reference track reduce cost?

Because it removes guesswork about direction. Words like clean, warm or powerful mean different things to different people, so without a reference the first version is an educated guess and the direction is often adjusted afterwards. A single linked track that sounds like your goal lets the engineer start in the right place, which typically cuts one full revision round out of the project.

YOUMIX (English) › What to send your engineer

What to Send Your Mixing Engineer

Published August 2026 · Guides for utaite, youtaite & VTuber covers

Written by the YOUMIX engineering teamTokyo, Japan — 3,000+ utaite & anime cover mixes delivered

Most delays in a mixing commission are not caused by difficult music. They are caused by files — the wrong format, effects that cannot be undone, or tracks that do not line up. Every one of these is avoidable in about ten minutes before you send anything.

This guide covers exactly what to send, in what format, and how to read the Japanese terminology you will run into if you commission an engineer in Japan. If you are working on a Japanese song cover and have not sorted out your instrumental yet, start with where to find official off-vocals first.

In this guide

The three files you always send
Japanese mixing terms, decoded
Export settings that actually matter
The single most common mistake
Why every file must start at the same point
Harmonies, doubles and ad-libs
Reference tracks — the cheapest thing you can send
Pre-send checklist

The three files you always send

#	What	Details
1	Your vocal, unprocessed	WAV, 24-bit, no effects of any kind. Harmonies and ad-libs as separate files.
2	The instrumental	The exact file you recorded against, at the best quality you have. If you only have an MP3, say so.
3	A reference	A link to a finished song that sounds like your goal. One is enough.

Anything beyond these three is a bonus, not a requirement. If you have a lyric sheet with the sections marked, or notes about which take you prefer at a particular line, send them — but do not delay the project gathering extras.

Japanese mixing terms, decoded

If you commission an engineer in Japan, some of these will appear in the brief. They are not interchangeable, and confusing two of them is the usual cause of a wasted round trip.

Term	Japanese	What it actually refers to
inst	インスト	The backing track with no lead vocal. What you sang along to.
off-vocal	オフボーカル	The same thing. Preferred term in Vocaloid circles.
dry	ドライ音源	Your voice with no effects at all. This is what you send.
2mix	2ミックス	Ambiguous. Can mean a stereo instrumental, or a rough mix of vocal plus instrumental. Always confirm which.
para	パラ	Separated tracks. In cover work it usually means your vocal parts split into individual files.
a cappella	アカペラ	Vocal only, no backing. Sometimes requested as a deliverable, not as a source.

The one to watch is 2mix. Japanese engineers do not use it consistently — some mean any finished stereo file, others specifically mean your rough mix. If it appears in a brief, one clarifying message costs you thirty seconds and can save a day.

Export settings that actually matter

Setting	Use	Why
Format	WAV	MP3 permanently discards high-frequency detail that de-essing and pitch correction depend on
Bit depth	24-bit	More headroom for correction. 16-bit works but gives away margin for nothing
Sample rate	Match your project (usually 48 kHz)	Converting adds a processing step with no benefit
Effects	None	Cannot be removed later. See below
Normalisation	Off	Changes level without improving anything, and can disguise clipping
Peak level	Roughly −12 to −6 dB	Comfortable. Anything touching 0 dB may already be clipped
Start point	Beginning of the project	Keeps every file aligned. See below
File names	Descriptive	`MainVocal.wav`, `Harmony_High.wav` — not `Audio 1.wav`

The single most common mistake

Sending vocals with effects already printed into them. It is the most frequent avoidable problem we see, and it is almost always well-intentioned — the singer added reverb because it sounded better while recording, then exported without removing it.

Here is why it matters. Mixing largely consists of shaping a raw signal: controlling dynamics, removing resonances, correcting pitch, then placing the voice in a space. Effects printed into the file sit in front of every one of those steps, and cannot be undone.

If this is printed in	What it prevents
Reverb / delay	The tail is baked into the sound. Any new reverb stacks on top, and the vocal cannot be moved forward in the mix
Compression	Dynamics are already flattened. Quiet detail cannot be recovered
Autotune / pitch correction	Artefacts are permanent, and further correction compounds them
EQ	Removed frequencies are gone. Boosting them back raises noise instead
Noise reduction	Aggressive settings thin the voice in ways that cannot be reversed

If you monitor with reverb while recording — and you should, it helps you sing better — make sure it is monitor-only. In most DAWs that means putting it on a send or an output bus rather than directly on the recording track. If you are unsure whether your export is clean, listen to it in isolation: if you hear the space, it is printed in.

If you genuinely want a specific effected character, send the clean take and a rough version with your effects as a reference. Then the sound is communicated without being locked in.

Why every file must start at the same point

Export every track from the very beginning of the project — timeline zero, or bar one — not from where the singing starts.

The reason is simple: files exported this way all have the same length, so they drop into any session and line up automatically. Files trimmed to their first note each have a different offset, and the engineer has to align them by ear. On a track with several harmony layers this is slow, and every manual alignment is a chance for a timing error.

Your DAW will call this something like “export from project start” or “whole session length”. Leading silence is not a problem — it is the point.

Harmonies, doubles and ad-libs

Keep them separate. One file per part, all starting at the same point.

The reason is control. Harmony sits behind the lead; doubles widen it; ad-libs need to appear and disappear. If they are bounced together into one file, none of those decisions can be made independently, and the engineer is left balancing choices you already made permanently.

A workable naming scheme:

01_MainVocal.wav
02_Harmony_High.wav
03_Harmony_Low.wav
04_Double.wav
05_Adlib.wav

Numbering matters more than it looks — it keeps the order intact when files are unzipped on a different operating system.

Reference tracks: the cheapest thing you can send

A reference is a finished, released song that sounds roughly like what you want. It is the highest-value item on the list, and it costs you a single link.

Descriptive words do not survive translation between people. Clean, warm, powerful, natural — every engineer has heard all four used to describe opposite things. A reference removes the ambiguity completely.

How to choose one well:

Pick a similar genre and vocal type. Referencing a heavily produced pop record for a quiet acoustic cover does not transfer.
Say what specifically you like about it. “How present the vocal is in the chorus” is far more useful than “this vibe”.
One is enough. Three references pointing in different directions is worse than none.

In practical terms this usually saves a revision round. Without a reference the first version is an informed guess, and direction is adjusted afterwards. With one, the first version starts in the right place. On a project with unlimited revisions it saves days; on a fixed-revision project it can save money.

Pre-send checklist

	Check	If not
1	All vocals exported as WAV, 24-bit	Re-export. Do not convert an MP3 to WAV — the loss already happened
2	No effects printed in	Bypass everything on the track and export again
3	Every file starts at project start	Re-export using whole-session length
4	Harmonies and ad-libs are separate files	Export each part individually
5	Instrumental is the exact file you sang to	Find it. A different master will not line up
6	Nothing peaks at 0 dB	Check for clipping and re-record if the take is distorted
7	Files are clearly named and numbered	Rename before zipping
8	One reference link included	Add one. It takes a minute and changes the outcome

Unsure whether your export is clean? Send one file and we will check it before you do the rest.
We mix utaite, youtaite and VTuber covers from Tokyo — all communication in English by email, X or Discord.

Request a mix

One last thing

None of this is about being technical for its own sake. Every item above exists to protect a decision that is still reversible. A clean 24-bit file starting at bar one keeps every option open; an effected MP3 trimmed to the first note has already closed most of them.

You only have to set this up once. After the first project the export settings become a preset, and you never think about it again.

Next: if you are still assembling the pieces of your cover, see the complete guide to making an utaite-style cover, or how to find official off-vocals for Japanese songs.