Recording a D&D session on Zoom, including the per-player audio trick

The free plan caps a meeting at 40 minutes

Zoom’s free Basic plan limits a meeting to 40 minutes and 100 participants. A D&D session is four hours. That means roughly six restarts a night, and — this is the part that matters — every restart ends the recording and starts a new one, so you come away with six separate files that each begin mid-conversation.

That is survivable. Six files transcribe fine and can be stitched back together afterwards. But it is worth knowing before you plan around it rather than at 11pm when you are trying to work out which file the boss fight is in. If you are going to run your game on Zoom every week, the paid tier removes the limit and is the difference between one clean recording and six ragged ones.

Computer recording, not cloud recording

Zoom has two recording modes and the distinction decides everything else on this page. Computer recording (Zoom’s current name for local recording) writes files to the machine you are sitting at. Cloud recording uploads to Zoom’s servers and is a paid feature.

Computer recording is the one you want, and not only because it is free. Zoom’s own documentation says computer recording "is available to both free and paid subscribers" and needs a "Basic (free) account or above". More importantly, the per-player audio option below is computer-recording only — cloud recording will not do it at any price.

One catch worth checking before you rely on it: computer recording is not supported on iOS, iPad or Android. Whoever is recording has to be on a desktop or laptop. If the DM runs the game from a tablet, hand the recording job to a player on a real computer.

Turn on per-player audio files before you need them

This is the setting that makes Zoom worth using for this. With it enabled, a computer recording produces one audio file per person instead of a single mixed track. Zoom names each one audio[Name][random number].m4a, so the file literally has the speaker’s name in it.

Why that is worth ten minutes of setup: transcription models do not label speakers. Feed them a mixed room and you get a wall of text with no idea who said what. Feed them one file per player and you can transcribe each separately, then label the output by filename — attribution you cannot otherwise get without paying for speaker diarization. It is the same trick that makes Craig’s multi-track output valuable on Discord.

The setting lives in the desktop client, not the web portal, and it applies to recordings made from that point on. It cannot be applied retroactively, which is the whole reason this section is near the top rather than the bottom.

1. Open the Zoom desktop client and go to Settings.
2. Open the Recording tab.
3. Tick "Record a separate audio file for each participant".
4. While you are there, note the recording location — Zoom defaults to a Zoom folder inside your Documents directory on Windows, Mac and Linux alike.
5. Record a two-minute test call with one other person and confirm you get two audio files, not one. Do this before session one, not after it.

What you actually get out

When the meeting ends, Zoom converts the recording and writes several files. The ones that matter for transcription are the audio.

The M4A is the file to transcribe if you did not enable per-player tracks. If you did, ignore it and use the per-name files instead — one transcription run each, then merge with the speaker names attached.

• video[random number].mp4 — the audio and video together.
• audio[random number].m4a — an audio-only file of the whole meeting.
• audio[Name][random number].m4a — one per participant, only if you enabled the setting above.
• playback[random number].m3u — a playlist for the MP4 files, Windows only.
• chat.txt — the in-meeting chat, which on a D&D table is often where the dice rolls and the URLs are.

Tell your players, and let Zoom tell them too

Zoom shows participants a recording consent dialogue, and its own usage reporting marks a player as having accepted it. That is a genuinely useful feature: it means nobody at the table can later say they did not know, and you have a record.

It is not a substitute for asking. The dialogue appears as a click-through in the middle of someone joining a game they are excited about, and everyone clicks it. Ask at session zero, say what the recording is for and how long you keep it, and let the dialogue be the backup rather than the conversation. The full version of that conversation, including why an online table spanning several states is the harder case, is in the consent guide.

Getting good audio out of a Zoom table

Zoom is built for meetings, which means it is aggressive about processing speech — noise suppression, automatic gain, and cutting the audio when it decides nobody is talking. For a meeting that is correct. For a game where someone does a gravelly goblin voice at half volume, it can clip the start of words.

The single biggest improvement is not a Zoom setting at all: everyone in headphones. Without them, each player’s microphone picks up every other player through their speakers, and the echo cancellation that fixes has to chew the audio to do it. Headphones remove the problem instead of processing it away. Beyond that, the same advice as any other table applies — a microphone closer to a mouth beats an expensive one further away, covered in the recording guide.

When Zoom is the wrong tool

If your table already lives on Discord, use Discord. A recording bot there gives you per-speaker tracks with no per-player setup and no 40-minute limit, and nobody has to join a second piece of software to play — the Discord workflow covers it.

If you play in person, none of this applies: one recorder in the middle of the table is the whole setup, and you will get one mixed track no matter what.

Zoom earns its place when your group is already on Zoom for other reasons, when you want video of faces rather than only voices, or when you specifically want the per-player files and do not want to add a bot to do it.