The domain adaptation component and the speech recognition models for the pharmaceutical domain resulting from the project will be implemented in the CLARIAH Infrastructure and thus be available for researchers in need of transcribing interviews (or more general, audio recordings) in this domain. Moreover, the project will deliver a methodology for adapting in-house speech recognisers for other domains where sensitive and special category data are involved. This opens the door to the use of (adapted) speech recognition by many organizations that can benefit from unlocking their, often sensitive, data. Examples are the Dutch Police (Tapped telephone conversations), psychiatric institutions (patient interviews), courts (court recordings), IND (interviews with immigrants), HRM departments (personnel interviews) and many and many more.

Via the portals that are being developed within the CLARIAH infrastructure such as the Media Suite hosted and maintained at NISV, researchers of various fields can make use of high quality speech transcripts in a variety of tools. These tools operate on the macro level (distant reading) and micro level (close reading) include fragment-level browsing (using time-codes available for every word in the transcripts), search (via the index of the transcripts), summarization (word cloud generation on the basis of the transcripts), entity-browsing (e.g., via named entity detection) and cross-media linking.  

Sandra: Nivel

The Nivel archive with thousands of AV-recordings of everyday medical visits in primary and secondary care can be analysed in a more time and cost-efficient way using the developed software. Consequently, policy measures resulting from the outcomes of the analyses will become more readily available. Other users, such as social scientists and healthcare providers working with the Nivel archive on dedicated projects, will also benefit from HoMed as it intends to elicit outcomes on a micro-analytic level which cannot be captured otherwise.

Toine: Freudenthal Institute

We would like to use the speech recognition software in our Pharmaceutical Humanities master course to help students to do research (transcribing interviews with patient and health care providers) in most efficient and reproducible ways.