BEGIN:VCALENDAR
VERSION:2.0
METHOD:PUBLISH
CALSCALE:GREGORIAN
PRODID:-//WordPress - MECv6.5.6//EN
X-ORIGINAL-URL:https://conference.eresearch.edu.au/
X-WR-CALNAME:
X-WR-CALDESC:eResearch Australasia Conference
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-PUBLISHED-TTL:PT1H
X-MS-OLK-FORCEINSPECTOROPEN:TRUE
BEGIN:VEVENT
CLASS:PUBLIC
UID:MEC-744124ddc0b2f3e512a0b7a7ee60b645@conference.eresearch.edu.au
DTSTART:20211014T030000Z
DTEND:20211014T032000Z
DTSTAMP:20210824T193900Z
CREATED:20210825
LAST-MODIFIED:20210914
PRIORITY:5
TRANSP:OPAQUE
SUMMARY:Developing an end-to-end audio transcription eResearch service in the Cloud
DESCRIPTION:Mr Smith A1, Mr Bennett R1, Haring M1, Ms Wyborn Y1, Mr Macintosh H1, Mr Williams C1, Prof. Bellgard M1\n1eResearch Office, Queensland University of Technology, Brisbane, Australia\nTranscription of audio files for researchers is an essential eResearch service in a number of research disciplines, especially in HASS. It is assumed that the content of these audio files is sensitive in nature meaning that researchers need to be careful of using external platforms to undertake the transcription process.\nThere are a number of challenges to be considered when using transcription services. Researchers need advanced features like speaker diarisation, dialogue-appropriate formatting and punctuation, and for transcriptions to be downloadable in a portable format. Institutions need to ensure that platforms are hosted in secure environments behind an organisational login, and that uploaded files have access-control, lifecycle policies, and examinable provenance. Considerations also need to be made for the both the horizontal and vertical scalability of services in order to meet unpredictable user demand and media resource requirements, while keeping costs low. This means that researchers must either use provisioned tools themselves, or request an eResearch officer oversee transcription on their behalf. Neither approach is scalable, and both cannot easily enable the researcher to undertake further downstream data analysis of the transcription.\nIn this presentation we outline an end-to-end ‘serverless’ solution that we developed where researchers can access with zero-friction using their institution credentials and receive prompt transcriptions of their uploaded recordings. We also highlight some planned features to empower researchers and improve usability such as interactive transcription alternative-selection and correction, domain-specific vocabulary and language models, and automatic personal identifiable information redaction.\n\nBiography:\nAdam Smith is a Software Developer with a passion for open-source, distributed systems, data analytics, and tinkering with code.\n
URL:https://conference.eresearch.edu.au/events/developing-an-end-to-end-audio-transcription-eresearch-service-in-the-cloud/
END:VEVENT
END:VCALENDAR