How to Set Up Translation Modes in NotchLive

October 8, 2026 · 6 min read · NotchLive 1.13.2

Read this guide in Chinese

Changing into Translate mode is only one part of setup. The language of the speech, the language you want to read, and how often the translation updates are separate choices. This guide shows where each lives in NotchLive 1.13.2.

For the broader workflow, see live translated captions on Mac. If you are still preparing permissions or the speech model, start with the main setup guide.

Check the requirements first

Built-in translation requires Pro, macOS 15 or later and a supported Apple Translation language pair. Prepare the speech model and the required language packs. NotchLive recognizes the speech first, then translates that text; neither update style guarantees instant output or a fixed delay.

App audio needs the macOS permission used for audio capture. Microphone speech needs microphone access, and Kōkako microphone captions also need the playback reference. The audio setup section explains those choices.

1. Choose Translate mode

Open Settings → Translation. Under Mode, select Translate. You can also change the mode from the live toolbar or notch control, or use the default ⌥⌘T shortcut. Mode and translation update style are different settings.

Choose your audio source separately: Apps & Media for speech playing on the Mac, My Voice for your microphone, or Everything for both. For example, a lecture playing in a browser usually starts with Apps & Media.

2. Set the spoken language and reading language

In Settings → Speech & Audio, set Apps & Media Language and Microphone Language for recognition. Auto is useful for changing languages. If the speech is mainly one language, selecting it explicitly can help recognition.

Then return to Settings → Translation → Translate To and choose the language you want to read. For an English lecture with Chinese captions, the source is English and the target is Chinese. Changing the target does not change what language the speaker is using.

3. Choose how live translation updates

Under Settings → Translation → Live translation, choose an update style:

• Faster updates: shows provisional translated text sooner. It may revise as speech continues and can use more battery. This is the default for new users.
• More context: waits for a completed caption before translating it, with fewer revisions and later output. Use the full Timeline caption window for this style.

The compact view and notch widget always use Faster updates. If you want to assess More context, open the full caption window, select Timeline and hide the notch widget. Regular speech recognition and saved-transcript translation are separate from this setting.

Try each style with the same short clip. Choose based on whether you prefer early text or steadier completed passages; translation accuracy still depends on the recognized speech and language pair.

4. Keep the original beside the translation

Enable Settings → Translation → Show original with translation. Bilingual captions give you the recognized original and its translation together, which helps when checking a name, number or phrase you partly understand.

The original is recognized text, so it can contain transcription errors too. For consequential wording, check the audio or ask the speaker rather than treating either line as a verified transcript.

5. Finish language-pack setup

In Settings → Translation → Translation Languages, find the required pair and choose Set Up…. The app may also offer this when a live translation needs a missing language.

1. In the macOS translation sheet, click every requested Download.
2. Wait for the downloads to finish, then choose Done.
3. Return to NotchLive and check the readiness state. Waiting for macOS is not the same as Installed.
4. If setup fails, use Retry. A pair marked Not Available needs a different supported choice; repeated downloads cannot add an unsupported pair.

The initial downloads need internet access. Once the model and language pair are ready, recognition and built-in translation run on your Mac without a NotchLive cloud translation service.

Translated Dictation has its own target

For text inserted into another Mac app, open Settings → Dictation → Translate dictation to. Choose its target and use its Set Up control if needed. This target is independent of the live-caption target.

With Dictation enabled and microphone and Accessibility permissions granted, hold the default ⇧⌥Space to use Translated Dictation. If you selected press-once behavior, use that instead. See the Dictation guide for insertion, dictionary and history controls.

A saved transcript uses language tabs

Open a saved session and its translation control. Set From and To, then choose Translate transcript or Pause captions & translate if the app needs to make room for the job. A text-only session can be translated; saved audio is not required.

Each translation gets a separate tab with its own search, copy, sharing and export. Original stays intact. You can leave the screen while the job runs; use Retry for a failed job or Cancel to stop it. The recording import guide covers the saved workflow.

One more setting during meetings

Meeting Intelligence has a separate Also auto-run in Translate mode option under Settings → Sessions & Meetings. It may slow translation. With it off, automatic analysis pauses in Translate and catches up in Caption; manual analysis remains possible. Meeting Intelligence additionally needs macOS 26 or later and available Apple Intelligence.