Speechmark

Speechmark captures your mic and the meeting audio on your Mac and keeps the recordings there. Nothing is uploaded for transcription, which is the core difference from cloud-first note takers. If your work involves confidential material, this is the reason to pick it.

The app splits speakers automatically and labels them once you confirm a name. After that, every transcript reads like a script with clear attribution. No more guessing who said what. That makes it far easier to scan who committed to what, especially in calls with several participants.

Rather than dumping bullet points, Speechmark writes the note you'd have written yourself: decisions, action items, and a short recap in plain prose. It's aimed at the meeting summary people actually read, not a transcript nobody opens.

You pick the model that fits your privacy bar. Apple Intelligence and a local Ollama model keep everything offline, while your own OpenAI or Anthropic key lets you use a stronger cloud model if you accept the trade-off. No single forced backend.

Every line in the note links back to the moment it was spoken. Click a decision and the audio jumps to that point, so you can verify a quote or catch a detail the summary compressed. It's a small feature that saves a lot of scrubbing.

Speechmark installs as a Claude extension through Settings, with no config files or terminal commands. Once connected, your transcript folder becomes memory Claude can reason over, and each answer cites the meeting it came from. Want to know what a client asked for three calls ago? Just ask, and Claude points to the exact meeting. The connector reads your notes locally and makes no network calls of its own.

There's no login and no meeting bot that announces itself in the participant list. You record from the menu bar and the note appears when you're done. For people who dislike bots joining calls, this removes the awkward part entirely. Simple as that.