Take part
Our voices, our data, our language
Kodava Thakk is run through the Kodagu.ai community — one open network, many projects. Join once, and take a seat in the language work below.
Voices of the corpus
Top contributors
Loading the honour roll…
Fresh from the hills
New recordings appear here the moment they arrive.
Ranked by time given to the corpus. Names appear as first name, okka, and village — exactly what each contributor consented to share.
How to communicate & contribute
Step 1
Join the community
Sign up on the Kodagu.ai community page — one form covers every project, and puts you in the member directory.
Join Kodagu.aiStep 2
Say what you can do
List yourself in the community directory with your role — recordist, transcriber, singer, developer — so the coordination team can reach you.
Add yourself to the directoryStep 3
Start contributing
Record your first clip today; everything else — sessions, transcription work, code — is coordinated by email and the community channels.
Record nowQuestions, partnerships, press: poonacha@cyberhuman.ai · Source code and issues: GitHub · The project listing on the hub: kodagu.ai/projects/kodava-thakk
Seats at the table
Every speaker
Record 30 seconds of Thakk on this site. Do it at the dinner table, at the ainmane, at the Kodava Samaja. Then get three relatives to do the same.
Give your voiceElder-session facilitators
The most urgent seat. Sit with the oldest fluent speakers of your okka and village — life histories, Palame, proverbs, place-name lore — and record the long versions. The team provides kits, training, and honoraria for elders.
Transcribers & validators
Paid, Karya-style work from home: listen to clips, write them in Kannada script per the project convention, or check others' transcriptions. Fluent listeners of every age are welcome.
Singers & tradition-bearers
Palame singers, wedding-speech specialists, dudikotpat drummers: your genres are the crown jewels of the corpus and are recorded with archival care, on video where you allow it.
Developers & ML engineers
The whole stack is open source — this site, the pipeline, and the coming ASR/TTS fine-tunes. Speech-model experience is gold; web and mobile hands are always needed.
Linguists & teachers
Own the orthography convention, design the evaluation sets, and shape the classroom pack with the Mangalore University MA program and the Kodava Sahitya Academy.
Diaspora organizers
Run recording drives at Kodava Samajas from Bengaluru to New Jersey; the contributor app works anywhere, and diaspora speech is a corpus stream of its own.
Funders & partners
Grants, CSR from companies with Kodagu ties, and diaspora endowment gifts fund kits, honoraria, and the paid transcription team. The corpus itself is the asset that keeps attracting support.
The rules we all play by
- Guardianship: the corpus is held in trust for the community. It is never sold; commercial use needs council approval and benefit-sharing.
- Consent is tiered and respected: archive-only, research, model training, public listening — every recording carries its speaker's choice, enforced in the pipeline.
- Credit, always: speakers, okkas, and villages are named wherever their voices are used (unless they prefer otherwise). Elders' voices are never cloned without family and council consent.
- Both dialects, equal dignity: Mendele and Kiggat are collected and evaluated on equal footing, and Kannada script and English travel together on everything public.