Forschungsgruppe II

European Social Science History Conference 2021: Transcribing Oral History interviews (semi-)automatically


Organisation eines Panels
Die Leiterin der Forschungsgruppe digital humanities, Almut Leh, hat mit Projektpartnern von ein Panel bei der ESSH 2021 organsisiert.

Wednesday 24 March 2021 14.30 - 15.45

Transcribing Oral History Interviews (Semi-)automatically. Technological Promises in Practical Testing

Organizer: Almut Leh

Chair: Linde Apel

Christoph Draxler: Web-based Tools for Transcribing and Segmenting Speech

The Bavarian Archive for Speech Signals (BAS) provides a number of web-based speech processing services such as online editors, multilingual automatic seg-mentation, anonymisation and subtitling. In my current research, I am looking at the potential benefits of ASR for transcription, and on determining on how to pre-dict ASR quality automatically

Herdis Kley: Preparing Transcripts in a Digital Research Environment for Interview Collections

The project Oral-History.Digital designs and implements an infrastructure for scientific collections of audio-visually recorded narrative interviews, providing a working environment with tools for transcription, alignment and indexing. Automatic alignment tools such as WebMAUS can be integrated to reduce the amount of time usually needed to link audio and text manually. Yet a profound preparation of the text file is still necessary to get better results.

Peter Kompiel: Iron Curtain - Using (Semi-)automatic Transcription Tools For Creating a Digital Interview Archive

For the creation of new interview collections at Freie Universität Berlin, the process of time-coded transcription should be as efficient and streamlined as possible. The project "Iron Curtain. Fatally failed escapes and the miscarriage of justice against prospective emigrants and refugees" uses an elaborate workflow reaching from speech recognition to the integration of a finished transcript in an archive software which can also serve as a model for other interview projects.

Almut Leh: Audiomining – Advanced Speech Technologies for Transcribing, Indexing and Analysing Oral History Interviews

Audiomining is a tool for analysing spoken speech with direct access to the audio signal. In addition to the conversion into text, timecodes are set with word accu-racy. This allowed the synchronous representation of audio/video and transcript in the form of subtitles. In the interpretation of oral history interviews, this can direct the focus from the text to the original source, the spoken word.

Zur Website der Tagung: