sayksii/Aria
ARIA - AI Realtime Intelligent Audio | Universal real-time AI subtitles for Windows
What it solves
ARIA is a Windows desktop application that converts system audio into live subtitles. It allows users to transcribe audio from sources like video players, browsers, games, and calls in real-time, providing a movable overlay for the recognized text and an optional separate overlay for translations.
How it works
The app captures Windows system audio or microphone input and processes it through one of three recognition modes:
- Precise: Uses Faster-Whisper for high-accuracy transcription of recorded videos or talks.
- Realtime: Uses Sherpa-ONNX or Vosk for fast updates during conversations and streams.
- Live Captions: Leverages the built-in Windows 11 Live Captions feature for everyday playback and meetings.
For translation, ARIA can use online services (Google Translate, Bing Translator, Youdao) or a local NLLB model for offline, predictable availability.
Who it’s for
Windows users who need real-time transcription and translation of their system audio for accessibility, language learning, or content consumption across various applications.
Highlights
- Multiple Recognition Engines: Supports Faster-Whisper, Sherpa-ONNX, Vosk, and Windows 11 Live Captions.
- Local Offline Models: The Full package includes offline models for transcription and translation (NLLB), ensuring privacy and privacy-focused operation.
- Flexible Overlays: Independent, movable and resizable subtitle and translation overlays.
- Multilingual Support: Broad language support across different modes, including Chinese, English, Japanese, and Korean.
- Embedded Python: Includes its own Python runtime, making it easier to install and run on Windows.
Related
- Project
- Project
- Project
- Project
- Project