Putting te reo artificial intelligence to the test

Te Hiku Media is a charitable media trust based in Kaitaia, with part of their work focusing on te reo revitalisation and enrichment. As part of this mahi, Te Hiku Media have developed an AI transcription tool for use in their broadcasting arm to help generate transcripts of their Te Reo recordings. This ever-evolving tool is performing very well for their purposes, providing their audience with an alternative way of exploring their output. Now, they are looking to see how the AI could further develop and be utilised in the GLAM sector by challenging it with different regional dialects, qualities of recording, and other modes of voice like mōteatea and waiata.

Recordings held in the Archive of Māori and Pacific Sound are ideal for such a collaboration.

This project will see a sampling of recordings put through the automatic transcription process, then the output compared against human-generated ones — gauging accuracy and completeness. Features and functionality of the tool will also be assessed, with results shared with the wider galleries, libraries, archives and museums sector for feedback and comment.

The key goal of the Archive is the long-term preservation of valuable information and knowledge, and to make sure they can be discovered and accessed by the communities, researchers, and individuals who would like to use them.

By working with this home-grown initiative, it is hoped that new methods of discovery may be created for this rich body of material, and that future development opportunities may be highlighted to further improve discovery and access.