How to prepare vocals for mixing: recording, takes and files

Recording: levels and signal chain

The most common defect is clipping. Digital clipping can’t be fixed in the mix, while a quiet but clean signal can be raised without loss.

• Level. Aim for peaks around −12…−6 dBFS. Don’t chase “full scale”: headroom saves you on loud phrases.
• Bit depth. Record at 24-bit — the dynamic range is enough that you don’t need to chase loudness. The sample rate is usually 44.1 or 48 kHz; what matters is that it matches the mix project.
• Signal chain. It’s best not to bake a limiter, heavy compression or reverb into the recorded track. Light processing while recording is fine, but if in doubt, record it clean.
• Room and microphone. Distance to the mic, a pop filter and no background noise (fridge, window, computer) do more than any plugin later.
• Monitoring. Check that your headphones don’t bleed into the microphone: the beat on the recorded track will be audible during mixing.

Takes and building the main version

Record several takes of each phrase and keep all of them — the engineer may need them to replace a word or to beef up the chorus. But so they don’t have to listen to all forty, build a main version:

1. Listen through the takes and mark the best ones by line or word. Most DAWs have a take-comping feature for this.
2. Assemble one main track from them — the “lead vocal”.
3. Put doubles, backing vocals, harmonies and ad-libs on separate tracks. Don’t mix them into the lead.
4. If you haven’t comped anything, put all the takes in a separate “Takes” folder and write in the notes which ones you like.

File format and where tracks start

• Uncompressed WAV (or FLAC), 24-bit, sample rate as in the project. MP3 and other compressed formats aren’t suitable for mixing.
• Mono for a mono source. One microphone — one mono track; don’t send a stereo file with two identical channels.
• One zero point. Export all tracks, including the beat, from the same start — the first bar of the project. Then they’ll line up in someone else’s project without nudging.
• No effects on export. Export the “dry” signal. If reverb or delay is important to the idea, send a separate version with the effect as an example.
• No normalizing or loudness boosting. Don’t raise the file to “maximum” — just export it as recorded.
• Check. Run the tracks through the analyzer: it finds clipping, DC offset and silence at the start.

How to name the files

A name should answer three questions: which song, what part, which take. For example:

• Song_Lead_Vox_v1.wav — the main vocal, first version;
• Song_Double_L.wav and Song_Double_R.wav — the double panned left and right;
• Song_Harm_High.wav — the upper harmony;
• Song_Adlib_Chorus.wav — ad-libs in the chorus.

Use Latin letters, digits and underscores: that way files don’t break when passed through messengers and cloud storage. Put everything in one folder and send it as an archive or a link.

Tuning and noise

• Tuning. There’s no single rule: some engineers want a raw vocal and tune it themselves, others want it with your rough correction. Ask in advance. If you couldn’t, send both: raw and corrected.
• Noise reduction. Don’t bake it into the file. Aggressive noise reduction leaves artefacts that can’t be removed later. Better send room tone — a few seconds of recording with no singing: the engineer will choose the processing from it.
• Gate. Export without a gate: it can chop off word starts and tails, and the engineer will have a choice.
• Breaths and sibilance. Don’t cut every breath out — it makes the vocal unnatural. Harsh “s” and “sh” sounds are fixed in the mix.
• Fades and clicks. Short fades at the start and end of clips remove clicks.

What to send to the mixing engineer

1. Vocal tracks: the lead vocal, doubles, backing vocals, ad-libs and, if needed, a “Takes” folder.
2. The beat or stems. If you bought the beat, send the stems (separate instrument tracks) or a full WAV, as agreed. What to check in the terms is in the guide “How to order your first beat”.
3. BPM and key. You can find them with the separate tool.
4. A reference. One or two tracks with a note on what you like in them: vocal level, space, brightness. A reference is a direction, not a demand to “make it identical”.
5. The lyrics — so nobody has to guess at unclear words.
6. Session notes (below).

Session notes: what to write

A short text file or an email. A few lines are enough:

• song title, artist name, session version, sample rate and bit depth of the files;
• BPM, key and any tempo changes;
• which takes are the main ones, where the problem spots are (“a click at 1:12”);
• what you want: “vocal up front”, “more air”, “wider chorus”;
• what must not be changed — for example, a special effect on a particular line;
• the deliverables you need: the main mix, an instrumental, an a cappella, a radio version (if needed);
• the deadline and the number of revision rounds you agreed on.

If you mix it yourself rather than sending the vocals out, the structure is the same — the notes will help you get back into the session a month from now. And if you need a finished result from start to end, take a look at the Stream-Q studio: it builds a mix from stems and masters it.

What to do after you get the mix

• Listen on different systems: headphones, a speaker, a phone, a car.
• Compare with the reference at the same loudness. The loudness of a mix and a master are different things — don’t judge by it.
• Write revisions by timecode, not “a bit quieter”: “0:45 — the vocal sinks a little in the verse”.
• Save the mix version you settled on and send it to mastering.