What’s new
What changed in each version for everyone, newest first. The app updates itself: when a version comes out, it offers to install it.
0.8.22
- The status window, the first-run wizard, the record offer and the update banner keep to one spacing scale.
- Models on disk in settings sit in a box like the rest of the tab.
- The wizard calls the speakers You and Other, as the transcript does; an empty model list now points to General.
- The status window puts a recording's name in quotes.
- In a recording's chat, Commands is a proper button, and each question and answer is labelled with who is speaking.
- Theme and Window language in settings line up.
0.8.17
- A recording can be renamed: choose Rename… in its menu or click its title on its page. Its folder on disk keeps its name.
- Save audio… under Share and in a recording's menu saves its sound as a separate file named after the recording.
- Click a phrase in the transcript to play from there; the phrase being spoken is highlighted.
- Transcribe with… is no longer offered for recordings without sound.
0.8.16
- What's new in the main window: after an update you can see what changed, including in versions you skipped. The main window opens when you click the Dock icon or choose Open Noum.ai in the menu.
- Calls no longer get quieter in AirPods and other headphones: echo cancellation turns on only while sound plays through the Mac's or a monitor's speakers.
- The Transcribe and Summarize buttons now sit in the middle of the tab, next to the progress. Pressing one right after a recording is no longer lost: transcription starts as soon as the audio is ready.
- Recordings without a folder now live under Recordings.
- If dictated text doesn't appear, the strip tells you to press ⌘V.
- The Start recording button in the status window no longer overlaps the switches.
0.8.15
Whisper transcription no longer writes words over silence: quiet stretches are not handed to it, the meeting language comes from Settings, and “Thank you.”, letters from another alphabet and phrases stretched over half a minute are dropped. Dictation: text that could not be pasted stays on the clipboard with a hint; instead of holding the key you can press it twice (the “How the key works” setting); if you like, a grey draft of your words shows in the bar while you speak; on a new Mac Parakeet Ultra dictates, and those who already used dictation keep v3.
0.8.14
Recordings play in both ears: before, without a call your voice was only in the left one, and on a call you and the other side were in different ones. Dictation no longer crashes the app if the audio device changes at the moment you let go of it, for example when AirPods connect.
0.8.13
In first-time setup, dictation is on when you reach its step, and the step’s main button is what asks for Accessibility.
0.8.12
First-time setup picks Summaries from Noumy by itself if nothing else on the Mac can write summaries, and shows it as a second option next to any Claude Code, Codex or Ollama it finds.
0.8.11
People on the other end of a call are told apart by voice, and names are matched more boldly. You can rename a speaker in the transcript with a click. The first transcription on a Mac downloads the voice separation model once, about 100 MB. Plus everything from the 0.8.10 beta: labels on the recording bar’s buttons, “Idea or bug…” and “Check for updates…” in the sidebar, a “recording started” banner that goes after 5 seconds, and Summaries from Noumy if you choose them.
0.8.8
Beta versions: Settings have a “Get beta versions” switch. When an update is ready, “Relaunch to update” appears at the bottom of the recordings window. The Russian interface has been proofread again.
Beta versions are not listed here: they go to those who turned on “Get beta versions” in Settings.
Try it on your next call
Installing takes a couple of minutes. After that the app keeps itself up to date.
Version 0.8.22, 28 MB, macOS 14.2 or laterHow to install