Designing Spontaneous Speech Search Interface for Historical Archives

Abstract

Spontaneous speech in the form of conversations, meetings, voice-mail, interviews, oral history, etc. is one of the most ubiquitous forms of human communication. Search engines providing access to such speech collections have the potential to better inform intelligence and make relevant data over vast audio/video archives available to users. This project presents a search user interface design supporting search tasks over a speech collection consisting of an historical archive with nearly 52,000 audiovisual testimonies of survivors and witnesses of the Holocaust and other genocides. The design incorporates faceted search, along with other UI elements like highlighted search items, tags, snippets, etc., to promote discovery and exploratory search. Two different designs have been created to support both manual and automated transcripts. Evaluation was performed using human subjects to measure accuracy in retrieving results, understanding user-perspective on the design elements, and ease of parsing information.