Language-data pipeline
Dataset export
The training corpus, built continuously from approved community annotations — Ede (demo language).
0.009
Hours of audio
4
Approved segments
3
Speakers
0
Community revisions
Licence & data sovereignty
This corpus is community-owned. It is held under a Kaitiakitanga-style guardianship licence: the community are the kaitiaki (guardians) of their language data and decide how it is used, shared, and by whom. Access to train models is granted by the community, not taken from it. Contributors are attributed; revisions are recorded transparently.
Seed text is English standing in for the target language — placeholder only.
Generated client-side from current session state — 4 approved segments.