UH Hilo Hawaiian language college turns to AI to help secure the future of ʻōlelo Hawaiʻi
The new collaborative Lauleo project aims to collect audio Hawaiian speech data to begin the process of creating AI to move Hawaiian voice to text. AND the project is being launched with a fun competition, open to all ʻōlelo Hawaiʻi speakers, cash prizes!
By Susan Enright/UH Hilo Stories.
The Hawaiian language college at the University of Hawaiʻi at Hilo is part of an international collaboration turning to artificial intelligence language tools to help secure the future of ʻōlelo Hawaiʻi (Hawaiian language).
The new Lauleo project aims to collect audio Hawaiian speech data to begin the process of creating AI to move ʻōlelo Hawaiʻi voice to text. The project is being done as a partnership between UH Hilo’s Ka Haka ʻUla O Keʻelikolani College of Hawaiian Language; Kanaeokana (of which Kamehameha Schools is a founding member), a statewide network of over 70 schools and organizations dedicated to strengthening the lāhui and nurturing the next generations of aloha ‘āina leaders; Awaiaulu, a group dedicated to developing resources and people that can bridge Hawaiian knowledge from the past to the present and future; and Te Hiku Media, a Māori language project based in New Zealand that spearheaded the effort and is now contributing from their experience toward ʻōlelo Hawaiʻi as a close member of the Polynesian New Zealand Māori language family.
“In the work of transcribing Native Hawaiian audio recordings from the online Kaniʻāina Hawaiian speech repository, it takes about 30 hours per one hour of speech to ascertain several drafts of transcription before it can be uploaded to share,” says Kumu Larry Kimura, a professor of Hawaiian language and Hawaiian studies at UH Hilo and a foremost authority on the audio archives of ʻōlelo Hawaiʻi. “Through this effort of Lauleo, the technology of speech to text will help to expedite the importance of audio information to promote the life of Hawaiian.”
- Learn more about Kumu Kimura’s digital work: UH Hilo digitizing 40-year-old ʻōlelo radio show; public can “listen to another time” (April 21, 2016, UH Hilo Stories)
With the motto, “Bringing our voices together for the future of our language: AI tools for ʻōlelo Hawaiʻi done by and for the community,” the Lauleo hui has announced that their project will take many contributing voices — I lau nā leo — to be a success.
Do you speak ʻōlelo Hawaiʻi? You are invited to contribute to the database anytime! Want to compete for cash prizes? Team up with some friends and contribute your voices between Feb. 23 to March 2!
The Lauleo project is being launched with a fun competition!
Lauleo is an ongoing, long-term project, but from Feb. 23 to March 2, there will be a massive voice data gathering effort in the form of a team competition with cash prizes.
Rules for the competition from the Te Hiku Media website:
Lauleo Reading Competition
February 23 – March 2
E hoʻomākaukau no ka hoʻokūkū!!!
- 4 or more people to a team (no upper limit)
- The app or website gives you sentences to read and records you reading them
- Each team must read 4 hours collectively to be in the running for large cash prizes (the app and website keep time for you)
- Teams have one week to read as much as they can
- Reviewers will listen to and approve sentences
- Good pronunciation is KEY
- Team rankings will be published daily
- First place = $5,000, Second place = $3,000, Third place = $2,000
This voice data gathering effort will help build the first speech-to-text tool for ʻōlelo Hawaiʻi, creating community-powered technology to support kanaka ʻōlelo Hawaiʻi of today and help our keiki to carry our language into the future.ʻO wai ana nā koa o ka heluhelu?!!!
Whether wanting to be part of the competition or part of the long-term effort, ʻōlelo Hawaiʻi speakers who would like to contribute to the database are asked to download the Lauleo app via Apple or Google Play or visit the Lauleo website; each are set up to record ʻōlelo Hawaiʻi speakers while reading specific sentences.
This data will be used to train a “model” that teaches the computer to hear ʻōlelo Hawaiʻi and turn what it hears into text. The ʻōlelo Hawaiʻi model can then be used to create tools for several things such as text messaging in ʻōlelo Hawaiʻi, creating an ʻōlelo Hawaiʻi Siri, or even building an app to help learners improve their pronunciation.
Anyone who speaks ʻōlelo Hawaiʻi can participate; participants do not need to be language experts, they just need a solid grasp of correct pronunciation and the willingness to do their best. Anyone from beginner to native speaker can join in the reading effort. “All of our voices are valuable and computers need to understand everyone if they are to serve a wide community of speakers and learners of various levels,” say the organizers in a statement about the project.
Why find support from New Zealand media for an ʻōlelo Hawaiʻi project?
Led by Māori language expert Peter Lucas Jones and Hawaiian engineer Keoni Mahelona (see photo at top of this post), Te Hiku Media is not a tech company run by billionaires, but a small, native-owned organization from the far north of Aotearoa working on the mandate of native speakers of their iwi to ensure the survival of te reo Māori (Māori language) into the future.
Organizers of the Lauleo project say those kūpuna saw what was coming and asked the younger generations to step up and shape a place for te reo Māori in the coming decades and beyond. “They recognized this as a form of tino rangatiratanga, or self-determination, and language sovereignty.” Te Hiku’s work began with transcribing hundreds of hours of archival audio of native speaking elders in their community, much of it generated by their iwi radio station, Te Hiku o Te Ika.
These recordings are “data” they hold in trust as kaitiaki (kiaʻi or guardians). While digitally preserving it for the coming generations, they are also using it to create language tools for te reo Māori. They understand all forms of Māori language data to be a taonga, or cultural treasure of immense value, to be used for the benefit of the language and those keeping it alive.
ʻŌlelo Hawaiʻi speech and text data contributed to the Lauleo campaign by individuals or institutions will be held and used by Te Hiku in the same way and with the same values. Language tools will be developed to serve the ʻōlelo Hawaiʻi community and their needs, both current and future.
“The hope is to collect speech data through the Lauleo reading competition and complete the first automatic speech to text tool in 2025,” organizers say in their statement.
This tool will be made affordable to the language community and revenue generated from subscriptions will be used to cover development costs and fund other digital needs for language revitalization, creating community-sustained support for ʻōlelo Hawaiʻi.
“The kūpuna left us an incredible legacy of language and it’s our job to steward it into the future,” state the organizers. “This means coming together to create quality technology and work toward true data sovereignty. Our cousins across the moana have forged a path in the AI world and are inviting us to join them because they know we are stronger together!” Click here to learn more about Te Hiku’s tools for Te Reo.
Get involved with the Lauleo campaign!
Spread the word! Follow the group on Instagram and Facebook @lauleo25. Download the Lauleo app from Apple or Google Play and start practicing reading. Encourage colleagues, students, friends and family to form teams and get ready for the Lauleo Competition. Share it with your hui kamaʻilio, hālau hula, hui hoe waʻa, and anywhere Hawaiian language speakers and learners might be.
Lauleo representatives will be at all the Ola Ka ʻĪ events in February, the public is invited to come and learn more and check out some demos of existing technology. “With your support we hope to rally as many voices as we can to make this happen. E lauleo kākou — Let’s all contribute our voices!”
Story by Susan Enright, a public information specialist for the Office of the Chancellor and editor of UH Hilo Stories.
