macOS AppTroubleshooting
No summary appears, system audio isn't captured, transcription falls behind, and other fixes.
No summary appears
Check that Apple Intelligence is on. By default, summaries use Apple Intelligence (Apple Foundation Models). Enable it under System Settings > Apple Intelligence & Siri and wait for the language model to finish downloading (right after a macOS update the model can take tens of minutes to get ready; waiting fixes it).
If Apple Intelligence can’t be enabled, for example on a company-managed Mac, Otonota says “Apple Intelligence isn’t available on this Mac” at launch and offers to download a local model that runs on the Mac (MLX Qwen3.5 2B, about 1.8 GB) for summaries instead. Choose “Download and use it” and summaries work in the free version. To switch later, click the engine name in the summary pane and pick Qwen3.5 2B.
Summaries update only once the window’s worth of segments has accumulated; nothing appears in the first moments. A smaller window shows results sooner.
System audio isn’t captured
- Check that the audio source is “system audio only”
- Check that Otonota is allowed under System Settings > Privacy & Security > Screen & System Audio Recording
- After changing the permission, relaunch both the meeting app and Otonota
The other side is captured, but not me
With “system audio only”, your own voice isn’t captured, because the meeting app doesn’t play it back. The current version can’t record both at once. To keep your own voice, switch to “microphone only”, skip the headphones, and let the other participants play through the speakers.
Transcription falls behind
If the transcript pane header shows a “Load”, “High load”, or “Audio skipped” badge, the engine isn’t keeping up with the audio. This is rare with the built-in engines but can happen when other apps are using the CPU or GPU.
- Close other heavy apps
- (Pro) With WhisperKit, switch to a smaller model
Recording itself never stops, and committed sentences are kept.
I need a language the selected engine doesn’t transcribe
The transcript language menu lists only the languages the selected engine supports. The Mac’s built-in engines cover 10 (Japanese, English, Chinese, Korean, Spanish, French, German, Italian, Portuguese, Russian). For any other language, click the engine name in the transcript pane, switch to WhisperKit large-v3 (included in the free version; 22 languages), then choose the language. The interface, summaries, and translation work in 22 languages.
The app crashed
The audio and transcript so far are saved every few seconds, so they are still there. On next launch, Otonota notes that the recording didn’t stop cleanly and asks whether to resume recording in the same session. If you do, a new recording block starts in that session; if you don’t, the saved data is still there to read. If it happens again with the same steps, use Help > Report a Problem…, which fills in your version and environment.
Glossary terms aren’t applied
- Check the format:
correct spelling(how it's said). A comma outside the parentheses ends the spelling there - Only the first 30 lines reach the summary and translation prompt; move important terms up
- Recognition hints are supported by the Mac’s built-in Apple Speech and similar engines, not by WhisperKit (spelling normalization after commit works with every engine)