The indigenous languages technology (ILT) project at NRC in 2021
summary
Project starts with $6M (2017) to build community-driven tools; code is open-source and governed by an all-Indigenous advisory committee. Deliverables include ReadAlong Studio that semi-automates audio–text alignment for educator videos, a verb conjugator for Kanyen’kéha plus frameworks reused for Michif and Algonquin, an Inuktut–English corpus of 1.3 million aligned sentence pairs for MT, and a speaker-dependent ASR workflow (train on ~2 hours, then auto-transcribe the remaining ~68) that speeds transcription; preliminary TTS for Mohawk sounds promising.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 0:20Social challenges
- 2:54Other challenges
- 4:07Language families
- 4:47Current situation
- 5:48The unfortunate history
- 6:46The NRC team
- 7:23Funding sources
- 7:57Readalongs
- 8:28Readalong Studio
- 11:45software
- 12:59Inuktitut
- 14:14Tools for translators
- 14:48Morphological analyzer
- 15:12Automatic speech recognition
- 17:09Sound hunters
- 18:03Language courses
- 18:45Future work
- 20:12Contact information
related talks
40mThe indigenous languages technology project at NRC: An empowerment approach to language
26mMachine Translation into Learning
45mTranscend Language Barriers with Custom Machine Learning APIs: um...
26mResponsible technology under digital colonialism
27mIndian Sign Language Recognition(ISLAR)
21mMultimodal Open Source at Kyutai, From Online Demos to On-Device
14mAI for a Billion Voices: Taking AI where there is Maximum Impact
09m90% of India Doesn't Speak English