Feature Requests

I am running macOS 15.6.1 on an M3 Max, after a recent update I now finally get transcription from my microphone, but I do not get any other audio transcribed (e.g. other speakers in a Zoom call). I cannot get any AI features to work, I keep getting an error "ERROR tauri_plugin_local_stt::events: server_start_failed: ServerAlreadyRunning", but I can see in activity monitor there is a tauri://localhost process running. This seems like a really cool project, and I'd consider paying for it, but given that transcription and AI features don't work this is unusable for me. Does it actually work for others?

When listening to a meeting auto detect various speakers and have a identification mechanism. So after a meeting if I rename Speaker_1 to Bob, next time Bob is automatically detected.

Automatically extract your tasks after a meeting and export them to your preferred task tool (Apple Reminders at the minimum)

Right now the custom vocabulary is just a single line input field in the settings which makes editing difficult. In my IT job there is a lot of technical jargon, expressions and abbreviations that I'd rather manage in an external file that I can then also version-control or at least in a text area inside Hyprnote that supports scrolling.

I would like to train the model to improve the accuracy of terms captured during the meeting. E.g. Hyprnote captured the term "T and M" and determined this to be TNM (Technical Network Management) wheeras it is "Time and Material" for a consultancy

In-situ, poor transcripts are ok as they save on system resources. Post-call, it'd be nice if hyprnote could take the audio file and run it through a more robust transcription service. For now, I feel like I have to hang onto the audio files as Base - English (and even Small - English) don't provide enough accuracy. Speaker diarisation / fingerprinting would go a long way, too.

Would love for their to be an option for a dark mode theme on the app

Pretty much what the title says—I would rather have the default storage for notes be a user-selectable folder of Markdown files than an opaque SQLite blob. This would help users find their notes using macOS Spotlight and standard UNIX tools, and would eliminate the need to have multiple "Integrations" just to sync notes.

It should be possible to integrate with Google Calendar and Google Meet so that when joining a scheduled meeting, a recording can automatically be started. The system should create a meeting note pre-populated with the meeting title and participant list from the calendar event.

Summary Let Hyprnote automatically recognise when a call starts and begin transcribing without me needing to click a “Record/Transcribe” button. This removes friction, reduces missed notes, and guarantees every relevant call is captured from the first hello. Similar behaviour exists in tools like Bloks.app , so there’s clear precedent and user demand. Problem Manual start is easy to forget—especially when jumping between back-to-backs, answering quick ad-hoc calls, or joining late. The result: partial transcripts, inconsistent notes, and more admin after the fact. Proposed behaviour Hyprnote detects the start of a Teams meeting and auto-starts transcription. It keeps transcribing until the meeting ends or I explicitly stop it. Works for scheduled and ad-hoc meetings. Resilient to late joins (starts as soon as I’m in the call) and re-joins (continues seamlessly). How it could work (implementation notes) Primary signal: Microsoft Graph / Teams meeting presence and join events (scheduled and ad-hoc). Secondary fallback: Local process/window detection for Teams with meeting state heuristics. Calendar assist: Use upcoming meeting join links and times to pre-arm auto-transcribe. Controls & settings Global toggle: Auto-transcribe Teams calls (On/Off). Quick controls: menu bar toggle + hotkey to pause/resume. Visual indicator: subtle banner when auto-transcription is active. Privacy & compliance Optional consent banner on start (“Transcription active via Hyprnote”). Respect tenant/org policies where recording/transcription is restricted. Success metrics % of Teams meetings with full-length transcripts (increase vs. manual baseline). Reduction in “missed transcription” incidents. Time saved per user per week (fewer manual actions). NPS/CSAT for meeting notes quality and reliability. Why this matters This turns Hyprnote into a trustworthy second brain for calls—always on, never forgetful, so users can focus on the conversation, not the buttons.