Designing computer-assisted and collaborative transcription methods
How current speech processing methods can be used to support transcription and how these techniques can be incorporated in community-based transcription workflows? Recent research have tried to find solution to automatically transcribe languages in low-resource settings, i.e. when data are too scarse to build efficient Automatic Speech Recognition (ASR) system. While progresses have been made in a wide range of different subtasks such as phones recognition, automatic segmentation or keyword spotting, these techniques are too often disconnected from the field.
Grounded in the Australian Northern Territory, the research aim to take advantage of existing speech processing technologies to design collaborative and community-based speech transcription methods.