Add project: AI Language Partner

[duct-tape2]

Basic info

Application name: AI Language Partner

Application repo link: https://github.com/duct-tape2/ai-language-partner

Application home link: https://duct-tape2.github.io/ai-language-partner/

Application description: Local-first mobile and browser-based Japanese speaking practice for Korean learners, using pre-authored dialogue banks and local speech recognition and synthesis.

Qualifications

• Free software with an online source repository. The project is MIT licensed.
• Using Python for a considerable part of its functionality. The FastAPI backend implements the API, authentication, content, local speech-provider adapters, and tests.
• Well-known, or at least prominently used in an identifiable niche. It serves the specific Korean-to-Japanese speaking-practice niche and already has three external contributors with merged pull requests.
• Maintained or otherwise demonstrably still functional on relevant platforms. The public repository has more than 180 commits, active CI, current documentation, and a working hosted demo.
• An application, not a library or framework. It ships a React Native client, browser demo, and FastAPI service.

Additional notability info

The project deliberately avoids a runtime LLM dependency in its core Daily Talk path. Learners use pre-authored dialogue banks with local-first STT/TTS options, which keeps deployments predictable and low-cost.

The repository currently has four human contributors, including three external contributors with merged PRs, plus a public contribution workflow for Korean, Japanese, documentation, accessibility, and backend work.

I ran the current Apatite source against this entry: normalize and render both complete successfully with the compatible ruamel.yaml<0.18 dependency. This PR keeps the source change focused; rendering on the current upstream head also incorporates unrelated link corrections already present in projects.yaml.