Translate, transcribe, dictate — output almost anywhere.

Type into any app (⌘⌥D)

Keyboard simulation drops the text wherever your cursor is — no copy-paste, no extra window. Works in every app. macOS asks once for the "Accessibility" permission so LLT may type for you.

Choose content & mode

Insert original only, translation only, or both. Pick "Final" (clean, completed sentences) or "Live" (types as you speak and self-corrects) — dictate in one language, get another typed out.

LocalAgreement streaming

A streaming algorithm that only commits words once they stop changing — bringing smooth, flicker-free live typing even to fully local, 100% offline Whisper.

Remote-start from your iPhone

With the free LLT Mobile app your iPhone becomes a wireless mic over Wi-Fi and can remote-start dictation or translation on the Mac — a reliable replacement for Continuity. Speak into the phone, watch the text appear on the Mac.

Overlay

Floating overlay at the bottom of your screen. Shows original text (small) and translation (large). Auto-hides after 6 seconds. Rounded corners, dark semi-transparent background.

Show Original Text

Display the original transcribed text above the translation in the overlay. See exactly what Whisper heard and how Mistral translated it.

Save Transcript

Save the full session as a timestamped TXT file in ~/Documents/LiveTranslator/. Original + translation for every segment. Also supports SRT subtitle export.

Text-to-Speech

Read translations aloud using macOS system speech. Useful when you can't look at the screen — the translation plays through your speakers or headphones.

Auto-Gain

Automatically adjusts input gain to maintain consistent audio levels for Whisper. Targets -14 dB RMS with smooth adaptation. Manual gain also available (0.5x–4x).

Auto-Start Backend

Optionally start the backend automatically when the app launches. Otherwise, start it manually from the control panel. Backend always stops cleanly on quit (SIGTERM → SIGINT → SIGKILL).

Whisper Model Selection

Choose from 5 Whisper models: tiny (~75 MB), base (~150 MB), small (~470 MB), medium (~1.5 GB), or large (~3 GB). Download any model on demand, switch anytime in Settings > Whisper, and delete unused models to save disk space. Tip: small is usually enough and fast. Set the source language explicitly — auto-detection can cause Whisper to hallucinate on short or quiet segments.

Smart Pause Detection

Intelligent voice activity detection that waits for natural pauses before sending audio to Whisper. Produces cleaner, more complete sentences instead of choppy fragments. Dramatically improves translation quality — especially for longer speech and interviews.

Network Access & LLT-Mobile

Enable network access (0.0.0.0) so other devices can connect to LLT over your local network. Use the free LLT-Mobile iOS app as a wireless microphone — or as a standalone audio source feeding into LLT via Wi-Fi. No cables, no Bluetooth pairing issues.

STT Engines

Choose your speech-to-text engine in Settings > Engines: Whisper (local, offline), Deepgram, AssemblyAI, Google Cloud (REST or gRPC Streaming), Microsoft Azure, or Vosk (self-hosted). VAD is set automatically per engine — streaming engines detect pauses server-side.

Parakeet — the default engine

Parakeet (NVIDIA TDT) is the new default speech-to-text engine: very fast, fully local on Apple Silicon, with continuous streaming. Fine-tune its timing in Settings > Parakeet — pause response time and the maximum sentence length before a line is committed — to balance speed against cleaner full sentences. Whisper stays available as the multilingual, language-lockable alternative.

vMix Title Output

Send translations directly to vMix as live title updates via the vMix Web API. Configure host, port, input number, and custom field names for translation and original text. The translated text appears in your vMix title generator — format and style it however you want inside vMix.

Fully portable: use your iPhone as a wireless mic, run LLT on a MacBook, and send translations to vMix over Wi-Fi — live interview translation without cables.

OBS & Browser-Source Overlay

LLT serves a transparent subtitle page on localhost. In OBS — or any app with a browser source like Streamlabs, XSplit or vMix — add a Browser source with the shown URL (e.g. http://localhost:8077/overlay). Translations appear live in your stream, in the same style as the on-screen overlay, with a fully transparent background.

VRChat Output (OSC)

Send subtitles straight into the VRChat chatbox via OSC, like Kikitan. Choose whether to send the original, the translation, or both, set host and port (default 127.0.0.1:9000), and use the test button plus an optional notification sound. Ideal for streamers and gamers who want live captions in VR.