PhrasePact

PhrasePact is a mobile + web app that helps endangered-language projects capture usable speech data by paying fluent speakers micro-stipends for short, structured recordings. Communities or researchers create “phrase packs” (e.g., greetings, market talk, kinship terms) with prompts and consent terms. Speakers record on their phones, optionally add translations, and get paid after basic quality checks. The system focuses on practical, high-frequency language, not academic perfection. It includes offline-first recording, automatic audio cleanup, and a lightweight review queue so a local coordinator can approve submissions. The output exports clean audio + metadata for archives, teaching materials, or future ASR/translation work. This is an AI + traditional app: AI helps with noise reduction, segmentation, and draft transcription where possible, but the workflow assumes many languages have weak or zero existing models.