01 — Before anything technical
Ask everyone first
Recording law varies by country and, in the United States, by state. Some places require only one participant to consent; others require everyone.
Discord's own terms also prohibit recording people without their knowledge, so a recording made secretly can cost you the account regardless of the local law.
The practical answer is simple and takes five seconds: say you are recording, say what it is for, and say it on the recording so it exists in the file.
If you are publishing it, get that agreement clearly rather than assuming silence is consent. Nobody minds being asked, and plenty of people mind being surprised.
02 — The usual route
Recorder bots
Discord bots such as Craig join a voice channel and record it. The better ones record each participant to a separate track, which is the single most valuable thing you can have for editing afterwards.
Separate tracks mean you can fix one person's level without touching the others, clean up a noisy connection individually, and cut crosstalk cleanly.
The limitation is that a bot records what arrives over Discord, which is compressed for real-time transmission. It is fine for a conversation and it is not the quality you would want for a published interview.
You also need permission to add a bot to the server, which on someone else's server is not always yours to give.
If you use a bot, choose one that records per-participant tracks. A single mixed file removes most of your editing options.
03 — On your own machine
OBS and system audio capture
OBS can capture Discord's audio along with your microphone. On Windows this works through Desktop Audio capture with no setup. On macOS you need OBS's application audio capture or a virtual device such as BlackHole.
Assign your microphone and Discord to different audio tracks in OBS's Advanced Audio Properties, and you get two tracks in the recording: you on one, everyone else on the other.
That is better than one mixed file and still not per-person, because everyone on the call arrives mixed together on the Discord track.
Your own voice will be at full local quality on its track, which is a real advantage if you are the host and talk the most.
04 — The best quality
Everyone records themselves
The double-ender is the method that produces genuinely good audio: everyone records their own microphone locally while the call happens, then sends you the file.
Nothing goes through Discord's compression, so every voice is at full quality, and a connection dropout during the call does not appear in the recording at all.
You align the tracks afterwards using a shared moment. Ask everyone to clap once at the start, then line up the claps in your editor.
The failure mode is human: a guest forgets to press record, records the wrong device, or cannot work out how to send a large file. Record a backup with a bot or OBS at the same time, always.
Always run a backup recording alongside a double-ender. The one time a guest's file does not arrive is the episode you needed.
05 — Picking one
Which method for which job
For a casual recording, a game session, or a reference copy of a meeting, a bot is easiest and the quality is fine.
For something you will publish, record locally. Either a double-ender with everyone recording themselves, or a purpose-built remote recording tool that does the same thing automatically.
That is what those tools exist for: each guest opens a link, their browser records them locally at full quality and uploads as they go, and nothing depends on them remembering anything.
Hilite does this and we make it, so weigh that accordingly. Whichever route you take, the principle is the same: the further your recording is from the internet connection, the better it sounds.
Record each guest at full quality
A link to open, local recording, separate tracks, nothing to install.
Each guest recorded locally, in their browser.
Try Hilite free.