HAPTIC INTERFACE FOR SEGMENTATION AND ANNOTATION OF AUDIOTEXTUAL CORPORA

Prof. Dr. Daniel D. Hromada for NFDI4Culture Digitalisation Barcamp (8.10.2021)

am Anfang war die Geschichte ...

Haptic Interface

lexical segmentation

[{"id":"w1","w":"Alle","start":"322","stop":"523"},{"id":"w2","w":"Kinder","start":"739","stop":"1355"},{"id":"w3","w":"sitzen","start":"1572","stop":"2189"},{"id":"w4","w":"im","start":"2405","stop":"2622"},{"id":"w5","w":"Flugzeug","start":"2839","stop":"3889"},{"id":"w7","w":"Außer","start":"4105","stop":"4739"},{"id":"w8","w":"Chantal","start":"4938","stop":"5988"},{"id":"w10","w":"die","start":"6205","stop":"6622"},{"id":"w11","w":"ist","start":"6839","stop":"7888"},{"id":"w12","w":"im","start":"8105","stop":"8922"},{"id":"w13","w":"freien","start":"9122","stop":"10172"},{"id":"w14","w":"Fall","start":"11459","stop":"11905"},{"id":"w15","w":". ","start":"11921","stop":"11938"}]

sublexical segmentation

[{"id":"w0","w":"die","start":"653","stop":"1426"},{"id":"w1","w":"Wasch","start":"1459","stop":"2493"},{"id":"w2","w":"_","start":"2510","stop":"2708"},{"id":"w3","w":"ma","start":"2725","stop":"3477"},{"id":"w4","w":"_","start":"3498","stop":"3781"},{"id":"w5","w":"schi","start":"3812","stop":"4582"},{"id":"w6","w":"_","start":"4614","stop":"4897"},{"id":"w7","w":"ne","start":"4913","stop":"5266"}]

song / dialect annotation

a little demo ???

 

we are currently extending the interface with OCR / region-of-interest manual selection functionality

summary

browser-based tool for fast & frugal segmentation & annotation of sonic corpora

associates segments of sonic stream to discrete symbolic sequences (e.g. labels)

JSON metadata stored in the header of ogg/opus file (potentially also coupled with a PNG file)

coupling between visual (graphemosymbolic) and sonic mediated by means of a haptic modality (e.g. "finger pointing")

useful there where other approaches (e.g. automatic speech recognition) would fail (dialects & idiolects, ancient languages, artistic production etc.)

uses Kastalia Knowledge Management System as a backend

... to be continued ...

UdK Rundgang (29.10 - 31.10)

Tabula Rasa (Urania Berlin, 30.10)

Digital Education Hackathon (https://fibel.digital/digieduhack, 9-10.11)

*

dh@udk-berlin.de

twitter.com/DigiEduBerlin

bildung.digital.udk-berlin.de