Large-scale processing, indexing and search system for Czech audio-visual cultural heritage archives

Abstract

This paper describes a complex system developed for processing, indexing and accessing data collected in large audio and audio-visual archives that make an important part of Czech cultural heritage. Recently, the system is being applied to the Czech Radio archive, namely to its oral history segment with more than 200.000 individual recordings covering almost ninety years of broadcasting in the Czech Republic and former Czechoslovakia. The ultimate goals are a) to transcribe a significant portion of the archive -with the support of speech, speaker and language recognition technology, b) index the transcriptions, and c) make the audio and text files fully searchable. So far, the system has processed and indexed over 75.000 spoken documents. Most of them come from the last two decades, but the recent demo collection includes also a series of presidential speeches since 1934. The full coverage of the archive should be available by the end of 2014.

Dealing with Bilingualism in Automatic Transcription of Historical Archive of Czech Radio

“…For the transcription task, we have been adapting and enhancing a largevocabulary continuous speech recognition (LVCSR) system developed previously in our lab. During the first two years of the 4-year project, we have implemented most of the required functionalities and utilized the system to process, transcribe and index more than 75.000 documents broadcast since 1993 to present [2]. That period did not pose a particular challenge for our research as we could employ the existing system trained for contemporary Czech.…”

Using Various Types of Multimedia Resources to Train System for Automatic Transcription of Czech Historical Oral Archives

“…For this purpose, we have adapted our previously developed large-vocabulary continuous speech recognition (LVCSR) system to deal with broadcast recordings in Czech and Slovak and designed modules for speech indexation and search. During the first 18 months of the project, we have processed about 75,000 audio files (with total duration of 30,000 hours) and created a demo version of the web service that allows for smart search in the transcribed data [7].…”