common-voice

Common Voice is an open-source, crowdsourced project by Mozilla that collects and validates voice recordings from diverse global speakers to build a free, publicly accessible dataset for training speech recognition models. It is designed for developers, researchers, and underserved language communities who need inclusive, non-proprietary voice data. The project is interesting because it democratizes voice AI, ensuring technology understands real-world accents, dialects, and languages rather than just corporate or dominant speech patterns.