-
dinome 0.3.1 Stable
msemes released this
2026-10-15 09:59:50 +00:00 | 11 commits to main since this release- Speakers recognized by voice: when Meet does not show who is speaking, for example during a screen share, dinome names those lines after the call by comparing voices with the people it already heard in the meeting. Only clear matches get a name, marked with a small wave icon. Identify speakers above a past meeting's transcript does the same; the setting is in Settings → Transcription. A 26 MB voice model is downloaded once.
- A shared screen in Meet is no longer taken for a participant ("… (Presentation)").
- A meeting's folder and delete buttons moved to the meeting list: hover a meeting for them, or right-click it for all its actions (open, folder, export, copy, delete). Delete also works on the meeting focused in the list, after asking.
- Deleting a meeting now removes it from the meeting list at once; it stayed listed until dinome restarted.
- When Check for updates in Settings finds a new version, the button to update now appears right there, next to it, and not only in the title bar. While it downloads, Settings shows the progress.
- For support: the extension popup's Details can record what Meet's page does for 30 seconds and save it as a file, when speakers are not recognised.
- Settings → About and the extension's popup say who makes dinome and where to ask for help: dinome.support@mariosem.es. About also notes that dinome was developed with AI coding assistance.
- After updating, reload the dinome extension on chrome://extensions and refresh open Meet tabs.
Downloads
-
dinome 0.3.0 Stable
2026-10-15 08:25:28 +00:00 | 12 commits to main since this release- Meeting summaries: a few sentences, then key points, decisions, tasks and open questions, in the meeting's language. Choose on the setup guide or in Settings → Summaries how they are written: on this computer (a small model, a one-time download of 1 to 5 GB, nothing leaves the computer), or with ChatGPT or Claude using your own API key and any of its models. Each meeting is summarized when it ends, or press Summarize in its new Summary tab. Summaries are part of the Markdown export.
- Six local summary models to choose from: Gemma 4 E4B (recommended), Gemma 4 E2B, Qwen 3.5 4B, Qwen 3.5 2B, Gemma 3 4B and Gemma 3 1B. Settings shows each one's download size, license and how well it did in our tests.
- Settings → Storage shows where the meetings, and the speech engine with its models, are kept and how much space they take. Change… moves them to another folder or drive, with progress; if a move fails, everything goes back where it was.
- When you join a Meet call and dinome is not taking notes, a card in the call offers to, so you do not forget. While it shows, one click on the dinome toolbar icon (or Alt+Shift+N) starts notes. Turn it off with Don't ask again on the card, or Ask when I join a call in the extension's popup.
- Choose the meeting's language before notes start, in that card or the extension's popup. dinome uses a downloaded speech model for that language, and the top bar shows which. Languages without a downloaded model cannot be chosen yet.
- Hover the motto next to dinome's name for what it means: dixi, notavi, memini is Latin for "I spoke, I noted, I remember", and their first syllables make the name.
- After updating, reload the dinome extension on chrome://extensions and refresh open Meet tabs.
Downloads
-
dinome 0.2.0 Stable
2026-10-15 05:24:38 +00:00 | 13 commits to main since this release- A new look: a dark or light indigo design with dinome's own title bar, softly moving light behind floating panels, colour avatars for every speaker, and animations throughout; the Chrome extension's popup matches. Choose System, Light or Dark in Settings, or switch with the sun/moon button in the title bar.
- The player's seek bar is the recording's waveform in the speakers' colours. Hovering it shows who says what at that moment, and the line being heard fills in as it is spoken.
- While a meeting is recorded, a live meter under the transcript shows the sound arriving, the others above the line and you below it. New lines fade in, and Jump to live returns to the newest one after scrolling up.
- Ctrl+K opens a quick search for meetings and actions, and ? lists all keyboard shortcuts. Ctrl+. hides the notes for a wider transcript.
- Up to 16 people in a meeting each get their own colour, used in the transcript, the people filter and the waveform.
- Copy a speaker's lines with the copy icon that appears when hovering them. Settings shows the speech models as cards with download progress.
- When a meeting started while dinome was open, the transcript did not always scroll along with new lines. It now follows from the first line.
- A people filter next to the search box shows only the people you tick, any combination of them. The list shows how much each person spoke; search covers only them, and the seek bar shows when they talked.
- Search the transcript with the box above it, or Ctrl+F. Capitals and Croatian diacritics do not matter, so "budzet" finds "budžet". Enter steps through the hits, and they show on the player's seek bar.
- A player under the transcript plays a finished meeting's recording. Click a line to hear it. While it plays, the line being heard is highlighted and the transcript scrolls along. The seek bar shows who speaks when, and Space plays and pauses.
- Speaker names no longer pick up notices from Meet's page. When Meet cropped your self-view, your own lines could be labelled "Others might still see your full video." After updating, reload the dinome extension on chrome://extensions.
Downloads
-
dinome 0.1.3 Stable
2026-10-14 17:39:27 +00:00 | 14 commits to main since this release- Until the speech engine is installed, dinome opens on the setup guide instead of the last meeting. The engine status in the top bar opens the guide too.
- Settings explains why the model list is empty before the speech engine is installed, and offers to install it.
Downloads
-
dinome 0.1.2 Stable
2026-10-14 16:39:16 +00:00 | 15 commits to main since this release- Settings can remove a downloaded speech model to free disk space. The model a live meeting uses stays until the meeting ends.
- The English model downloads from Settings like the Croatian ones, with progress. Downloading a model no longer switches the model in use, so it is safe during a meeting.
Downloads
-
dinome 0.1.1 Stable
2026-10-14 15:55:10 +00:00 | 16 commits to main since this release- The installer, the app, its shortcuts and the taskbar now show the dinome icon instead of Windows' generic program icon.
Downloads
-
dinome 0.1.0 Stable
2026-10-14 15:32:08 +00:00 | 16 commits to main since this releaseFirst release.
- Live transcripts of Google Meet calls in Croatian and English. Speaker names come from Meet, and your own voice comes from your microphone.
- Speech recognition runs on your computer, on an NVIDIA GPU when there is one. The default Croatian model is ParlaSpeech Whisper turbo; BuzzASR Croatian and Whisper large-v3-turbo for English are alternatives.
- Markdown notes next to the transcript. Export or copy both as Markdown.
- Rename speakers, and re-transcribe a past meeting with another model.
- Windows installer. The speech engine and the models are set up on first start.
- dinome checks for new versions, asks before downloading one, and restarts into it only when you say so.
Downloads