RTranslator
An open-source Android app for real-time, offline voice translation between multiple languages using on-device AI models.
What is RTranslator?
RTranslator is an open-source Android application that provides real-time voice translation between multiple languages entirely offline. It uses on-device AI models like Meta's NLLB for translation and OpenAI's Whisper for speech recognition to enable conversations as if both parties spoke the same language. The app operates without an internet connection, ensuring user privacy and functionality in any environment.
Android users who need to communicate across language barriers in real-time, especially travelers, multilingual teams, or individuals in areas with limited internet access who prioritize privacy.
Developers choose RTranslator for its completely offline operation, strong privacy guarantees, and use of cutting-edge, optimized AI models that deliver high-quality translations without relying on cloud services or subscriptions.
Overview
Open source real-time translation app for Android that runs locally
Use Cases
Best For
- Travelers needing real-time conversation translation without internet access
- Privacy-conscious users who want translation without data sent to external servers
- Multilingual teams conducting meetings or collaborations across languages
- Quick on-the-street interactions like asking for directions or shopping assistance
- Educational settings for language practice or cross-cultural communication
- Offline environments where cloud-based translation services are unavailable
Not Ideal For
- Users with Android devices having less than 6GB of RAM who need reliable real-time translation
- Situations requiring instant setup without a large initial download, such as emergency or impromptu travel
- Projects needing high-quality voice translation for languages only available in low-quality or text-only mode, like Akan or Dzongkha
- Environments where Bluetooth connectivity is unreliable or not feasible for multi-device conversation mode
Pros & Cons
Pros
All translation and speech recognition models run locally with no internet needed after the initial 1.2GB download, enabling use in any environment without data concerns.
The app collects no personal data and processes everything on-device, as confirmed in the privacy policy and README, ensuring user data never leaves the phone.
Developer has quantized and separated NLLB and Whisper models to reduce RAM use by up to 1.9x and speed up execution by 4x, as detailed in the performance tables in the README.
Offers Conversation mode for bidirectional talks via Bluetooth, WalkieTalkie for quick exchanges on one phone, and Text translation, covering various real-world scenarios.
Cons
Requires at least 6GB of RAM and a fast CPU, which excludes many budget Android phones and can lead to crashes or poor performance on weaker devices, as stated in the performance requirements.
Must download 1.2GB of models initially, and in regions with slow GitHub, manual sideloading is needed, making installation less straightforward than typical app store downloads.
Speech output quality varies based on the phone's text-to-speech engine, requiring additional setup like installing Google TTS for best results, and troubleshooting may be needed if TTS fails.
As a beta app, it has known issues like Bluetooth connection drops, which can interrupt real-time conversations and affect reliability, as mentioned in the bugs section.
Frequently Asked Questions
Related Projects
Copy Translator🔠Foreign language reading and translation assistant based on copy and translate.
LibreTranslateFree and Open Source Machine Translation API. Self-hosted, offline capable and easy to setup.
Crow TranslateA simple and lightweight translator that allows you to translate and speak text using Google, Yandex Bing, LibreTranslate and Lingva.
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.