Using Otter.ai for D&D sessions, and the point where it stops

The numbers, with sources

How we sourced this: The plan figures above were re-read from Otter’s own pricing page on 2026-09-07. The conversation-history row comes from Otter’s help centre, which returns 403 to automated requests — that line was read from a search engine’s index of Otter’s article on 2026-08-05 rather than fetched directly, so treat it as the weakest row in the table and check it yourself if you are deciding on it.

The length problem, in numbers

A session is the whole evening. Ours run about four hours including the part where nobody can decide what to buy in town. Otter’s caps are per conversation, not per month, so the monthly minute allowance is not the binding constraint — the per-conversation ceiling is.

On the free plan that ceiling is 30 minutes. On Pro it is 90. Neither is a session. The first tier that covers four hours in one recording is Business, at $30 per user per month billed monthly. For comparison, one DM Scribe upload takes a session up to 8 hours and 500 MB whole — recordings over four hours are split and reassembled automatically on our side, not yours.

• Free: 30 minutes per conversation — roughly the first fight
• Pro: 90 minutes per conversation — about half a session
• Business: up to 4 hours per meeting — a session, just about

The workaround, and what it costs you

You can split the recording and upload it in pieces, and people do. Two things make that worse than it sounds. The free plan allows three audio or video file imports for the life of the account, not three per month, so splitting a session into chunks spends your entire import allowance on one evening. Pro raises that to ten a month, which a weekly game in 90-minute pieces will use most of.

The second cost is that the seams land mid-scene. A transcript broken into arbitrary chunks is still searchable, but anything that reads across the whole session — a recap, a summary, a question about what happened — now has to be done per chunk and stitched by you.

Names are the other half of it

Meeting transcription is tuned for meetings, where the proper nouns are ordinary names and company jargon. A campaign is mostly invented words, and a general model has never seen any of them. The same NPC comes back spelled three ways across three sessions, and the archive stops being searchable at exactly the point where it would start being useful.

This is the reason DM Scribe exists as its own product rather than a complaint about Otter. Before your audio is transcribed, we assemble a short vocabulary from the campaign itself — glossary terms with their pronunciations, the player characters, the nicknames the table actually uses — and hand it to the model as context, so Thordak arrives as Thordak in session nine because session eight taught us to expect him. See how that works.

What Otter does that we do not

Otter joins live meetings on Zoom, Teams and Meet and takes notes in the room, which is a genuinely different product from uploading a file afterwards. It has a mobile app built for recording in person. If you also sit in work meetings all week, one subscription covering both is a real argument, and we do not have an answer to it.