Remote collaboration: how to send files and stems properly

Remote music collaboration has made it normal to write with a producer in another city and record a singer in another country. What usually goes wrong is not the music. It is a vocal that starts two bars late, a session at the wrong sample rate, or "final_new2.wav" arriving with no idea which take it is. A few habits fix almost all of it.

Agree on the basics before anyone records

Ten minutes of agreement at the start saves days later. Settle these in writing:

• Sample rate and bit depth. Pick one for the whole project, for example 48 kHz at 24-bit, and make sure everyone's session matches.
• Tempo, time signature and key. The exact tempo, including decimals, and any tempo or key changes with the bar where they happen.
• The start point. Every file in the project starts at the same zero point, usually the very beginning of the session, even if the part enters much later.
• Who does what, and by when. Who records which part, who mixes, when each delivery is due.

Different music software can't open each other's projects, so audio files are the common language. For parts someone may want to re-voice, such as synths or drums, also send MIDI files.

Time zones matter more than people expect. Agree on when each of you is reachable, and if a part needs live direction, a singer's phrasing for example, book a video call for the recording session itself. Half an hour of real-time feedback can replace a week of messages.

Sending a guide others can record to

The person recording a part at home needs more than an idea. Send them:

• a guide mix of the current version
• a click track as a separate file, starting from zero
• a MIDI file or tempo map if the tempo changes
• the lyrics with the sections marked, or a chord chart
• one or two reference songs and a sentence about the feel you want

Picture a made-up producer, Ella, in Toronto, and a singer, Sami, in Lisbon. If Ella sends only a rough MP3, Sami will record to something that may not line up with her session at all. If she sends a guide, a click and the lyrics with bar numbers, Sami's vocal drops into place on the first try.

Recording your part and sending it back

When you are the one recording:

• Record to the guide in a session set to the agreed sample rate and tempo.
• Export only your parts, each starting at zero and running to the end of the song.
• Send them dry, without reverb or delay. If an effect is part of the idea, send a second, processed version as well and label it clearly.
• Don't tune, compress or clean the recording heavily unless that was agreed. It is easy to add processing later and impossible to remove it.
• Send your best comp plus the alternative takes, labelled so they can be found.
• Before sending, open the exported files in an empty session with the guide and check that they line up.

Names, versions and notes

A good file name answers three questions: which song, which part, which version. For example "Yay gecesi_Lead vocal_Sami_v03". Avoid the word "final"; there is always one more version.

Number every bounce and never overwrite an old one. Put a short text note in each delivery: tempo, key, sample rate, what is inside and what changed since the last version.

Feedback works best with timestamps: "1:42, second chorus, bass too loud". Collect all your notes for one round into a single list instead of sending them one by one, and keep the conversation in one place. Notes scattered across five chats are notes someone will miss.

Moving the files

• Use a shared cloud folder with a fixed structure. For example: reference, files from each person, mixes, notes. Everyone knows where to look without asking.
• Keep WAV as WAV. Convert to MP3 only for quick listening on a phone, and label those files as previews.
• Be careful with messengers. Many apps compress audio unless you send it as a file, and some links expire after a while. A cloud link is safer.
• Zip a delivery so that nothing goes missing on the way, and keep your own copy until the other person confirms that everything opened.

Before you hit send

1. The sample rate and bit depth match what you agreed.
2. Every file starts at zero and lines up with the guide.
3. Each name includes the song, the part, the person and the version.
4. Dry and processed versions are clearly labelled.
5. A text note with the tempo, key and changes is included.
6. You listened to the exported files, not just the session.

One more thing that isn't technical: write down who wrote and played what as you go. When the song is finished, that list makes credits and splits a short conversation instead of an argument.