About BabelDuck
BabelDuck is an AI conversation practice application for language learners, designed to be approachable for beginners and to reduce the effort and anxiety of practising spoken expression. Learners hold conversations with an AI partner and can ask for help at any point without breaking the flow of the exchange.
It is built for language learners at every level, with particular attention to those just starting out. While a conversation is under way, the learner can request grammar checks, translations or suggestions for more natural phrasing, using customisable quick commands. If a suggestion raises a question, a sub conversation can be opened to discuss it and the learner then returns to the original conversation where they left off. It supports voice input and spoken responses, multiple conversations with custom system prompts and streaming replies, and per conversation preference settings. Several LLM services are integrated and can be switched between, the interface is available in multiple languages, and built in tutorials guide new users. Conversation data is stored locally for privacy.
The hosted web app is available at the project website and the site publishes no pricing or paid plans. The source code is on GitHub, and the application can be self hosted with Docker by cloning the repository, filling in an environment file and running the published container image.
The code is released under a Time Limited Competition Protection Licence that converts to the AGPL in future, rather than a standard open source licence, so commercial reuse should be checked against its terms.