Speaker labels are the feature
Here are two transcripts of the same meeting.
The first one is a wall of text. Forty minutes of conversation, punctuation, the occasional "yeah, exactly" — all of it true, none of it useful. You can read it. You can't use it. To find the one thing you're looking for, you have to re-live the whole meeting.
The second one has names on it. Jeff said the vendor deadline was soft. Bob disagreed. You committed to sending the revised numbers by Friday. Same forty minutes, same words — but now it's a record you can act on.
That second transcript is the whole product. The names are what turn a recording into notes.
What names buy you
Two things, and both of them are the reason you record a meeting in the first place.
You can search it. A month from now, someone emails you: "wait, did Jeff actually agree to the soft deadline, or did I imagine that?" Without names, you're scrubbing audio. With names, you search "Jeff" + "deadline" and read the exact line, timestamp and all. The transcript stops being an archive you never open and becomes something you check the way you'd check email.
You can assign it. An action item without an owner is a wish. "Someone will follow up with the vendor" is the kind of line that dies in a summary — technically captured, functionally lost. "Bob follows up with the vendor by Wednesday" is a task. The only difference between those two sentences is a name, and the name only exists because the transcript knew who was talking.
Every downstream thing you want from a meeting — the follow-up email, the decision log, the "who owns this" — is built on knowing who said what. Strip the names out and you're left with a text file that proves a meeting happened without telling you anything you can do about it.
Why almost nobody does it
If speaker labels are so obviously the point, why do so many tools ship without them?
Because it's the hard part. Turning speech into text is close to solved — the models are good, they run on your laptop, and most meeting tools reach for the same handful. But figuring out who is speaking, moment to moment, from a single mixed audio stream where people talk over each other and the room echoes — that's a different and much nastier problem. It has a name, "speaker diarization," and it's where the engineering lives.
So it gets skipped, or deferred, or fenced off behind a paywall. The most popular open-source meeting tool on the internet right now — fifteen thousand stars, real momentum — is a good example. I looked at it closely this week because more than one person asked me if it made Fieldnotes redundant. It transcribes well. It runs locally. And it does not label speakers. Diarization is listed as a someday-feature for a paid tier that, as I write this, hasn't shipped in a single release.
I'm not knocking it — that's an honest reflection of how hard the problem is. But it tells you something that the most popular free meeting tool on the internet does the easy 80% and leaves the part that makes notes useful for "later." The easy part is table stakes now. The names are what's worth paying for.
What Fieldnotes does
Fieldnotes labels every speaker, on your device, today. It's free, and there's no upgrade that unlocks it later.
It records the mic and the system audio, transcribes the whole thing locally, and then — the part that matters — separates the voices and puts a label on each one. Out of the box you get "Speaker 1," "Speaker 2." You rename them once, and every line that person spoke updates at the same time.
And it's learned to recognize you. Fieldnotes builds a voiceprint from your own recordings, so the next time you sit down for a call, it pre-fills your name on your own lines without you lifting a finger. The tedious part of labeling — "which one of these is me again" — mostly disappears.
The privacy story is the same one I always tell, because it's the same architecture: none of this leaves your machine. The audio, the transcript, the speaker labels — all of it is computed on your laptop and written to your disk. There's no server that hears your meetings. Most tools can't say that, and the ones that can mostly can't tell you who was talking.
The tell
Here's the way I'd put it to anyone deciding between meeting tools.
Watch what a tool treats as the finish line. If the demo ends at "look, it made a transcript," ask who's talking. If it can't say — or it can, but only if you upgrade — then you've found the seam. The transcript is the part that's easy to show off. The names are the part that's hard to build, and they're the part you'll use every single day after the demo is over.
Notes without names are a wall of text. Notes with names are a record you can act on.
That's the reason to record at all.
— Ryan