Back to blog

Mac live caption translation has no audio? Check permissions first

Separate Mac system audio from microphone input and troubleshoot missing captions, recognition errors, and live translation failures step by step.

Published September 27, 2026

When live translated captions show no text, the translation model may not be the problem. macOS must first allow the audio to be captured, speech recognition must turn that audio into text, and only then can translation begin. Treating all three stages as one issue often leads people to reinstall the app or change languages without fixing the missing input.

This guide first separates audio playing on the Mac from sound entering through a microphone, then follows the processing chain from permissions to transcription quality.

Identify which sound you want to translate

The two common input paths on a Mac are:

  • System audio: sound from a browser video, podcast, media player, or meeting app that the Mac sends to speakers or headphones.
  • Microphone audio: speech picked up by the built-in microphone, a USB microphone, or an audio interface.

Apple's built-in Live Captions makes the same distinction between Computer Audio and Microphone and lets users switch sources. That is a useful mental model for third-party caption tools, but Apple Live Captions and a third-party translation app are not the same feature. Their permissions, language support, and processing methods should not be assumed to match.

Trans2 starts live subtitle translation with ⌥A by default. Its product page describes a pipeline that captures system audio, transcribes it with Speech, and shows the translation in a draggable caption window. If the sound you need comes from an in-person conversation or external microphone, first confirm that the current tool and workflow support that source.

Check system-audio access when media is silent

Apple places screen and system-audio authorization under Privacy & Security. Check it in this order:

  1. Open System Settings.
  2. Select Privacy & Security.
  3. Open Screen & System Audio Recording.
  4. Find the relevant app and allow it to record system audio.

Apple's documentation says macOS can allow an app to record both the screen and audio or audio only. A live-caption feature needing sound therefore does not automatically mean that the app must receive the picture on your screen. Use the narrowest option that macOS offers and the app actually needs.

After changing the permission, quit the app completely, reopen it, and test with a short video that definitely contains speech. Also verify that the player is not muted, the output has not switched to another headset or display, and the Mac is producing audible sound.

Check microphone access for people speaking nearby

If the source is you, a room conversation, or an external audio device, inspect a different permission:

  1. Open System Settings > Privacy & Security.
  2. Select Microphone.
  3. Enable access for the app that needs the input.
  4. Quit and reopen the app before testing again.

Apple notes that without microphone or audio-input permission, an app may not hear sound and may show no waveform or meter activity. The recording indicator in Control Center shows that the microphone is being or was recently used, but the indicator alone does not prove that speech is being recognized correctly.

If a USB microphone or audio interface still produces no input, also check the selected device and input level in Sound settings, then review the manufacturer's software and compatibility information.

Diagnose the pipeline in three layers

Once permissions are correct, continue in the order the data is processed.

Layer 1: audio capture

Play a short sample with one clear speaker and little background music. If there is no text or input activity at all, return to permissions, source selection, mute state, and audio-device routing.

Layer 2: speech recognition

If text appears but omits or changes many words, audio is probably reaching the app and the more likely problem is signal quality or recognition. Try to:

  • Reduce background music and room noise.
  • Avoid overlapping speakers.
  • Adjust playback volume without causing distortion.
  • Match the recognition language to the speech when the app offers that setting.
  • Compare with a clean sample containing one clearly spoken voice.

Apple also warns that the accuracy of its Live Captions can vary and that it should not be the only source relied on in high-risk or emergency situations. Third-party live transcription likewise needs human review, especially for numbers, names, medication, legal terms, or financial information.

Layer 3: translation

If the source transcript is mostly correct but the translation is weak, then inspect the target language, terminology, available context, and sentence boundaries. Adjusting translation settings cannot restore speech that was never captured or was transcribed incorrectly.

A two-minute troubleshooting checklist

This order is usually more useful than reinstalling the app:

  1. Decide whether the source is system playback or a microphone.
  2. Enable the corresponding system-audio or microphone permission.
  3. Quit and reopen the caption app.
  4. Verify the Mac's selected input or output device and mute state.
  5. Test with ten to twenty seconds of clean single-speaker audio.
  6. Check for a source transcript first, judge recognition second, and inspect translation last.
  7. Turn off audio permissions that are no longer needed after testing.

Permission and privacy are separate questions

A macOS permission only says whether an app may access a class of audio. It does not explain where the audio is processed, how long it is retained, or whether it is used for another purpose. Apple recommends reviewing a third party's privacy practices before allowing it to record the screen, system audio, or microphone.

With any live-caption tool, read the current privacy information and avoid unnecessary capture around customer data, account details, or confidential meetings. For Trans2's actual processing boundaries, rely on its current privacy page and the prompts shown by the installed version.

Sources

Related articles

Continue with articles that share the same product tags.

Mac screenshot OCR translation: improve recognition before judging the translation

Learn how OCR errors flow into screenshot translations, with practical ways to improve capture quality, language handling, terminology checks, and review.

How to choose the right BitBear product

Choose the right BitBear tool for Telegram translation, AI support, general AI, Mac translation, Telegram digital goods, or market research.