Interactive search for words and phrases in large audio-visual archives

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F49777513%3A23520%2F17%3A43932995" target="_blank" >RIV/49777513:23520/17:43932995 - isvavai.cz</a>

Výsledek na webu

—

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Interactive search for words and phrases in large audio-visual archives

Popis výsledku v původním jazyce

This paper describes an automatic system for processing and searching for large audio-visual archives, especially for the use of in the field of oral history studies. The system contains automated processing pipeline for speech recognition and indexation. The carefully designed graphical user interface allows to search for specific words and phrases. It also allows to search for out-of-vocabulary words and directly replay the occurrences sorted according to the automatically estimated confidence scores. The first archive processed in this system is the MALACH archive containing personal testimonies of holocaust survivors and witnesses. The searchable portion of interviews consists of 2,000 hours of English recordings and 1,000 hours of Czech recordings. The paper gives a brief overview of the architecture of the system, describes the phoneme-based search and provides basic performance metrics for both the English and Czech data.

Název v anglickém jazyce

Interactive search for words and phrases in large audio-visual archives

Popis výsledku anglicky

This paper describes an automatic system for processing and searching for large audio-visual archives, especially for the use of in the field of oral history studies. The system contains automated processing pipeline for speech recognition and indexation. The carefully designed graphical user interface allows to search for specific words and phrases. It also allows to search for out-of-vocabulary words and directly replay the occurrences sorted according to the automatically estimated confidence scores. The first archive processed in this system is the MALACH archive containing personal testimonies of holocaust survivors and witnesses. The searchable portion of interviews consists of 2,000 hours of English recordings and 1,000 hours of Czech recordings. The paper gives a brief overview of the architecture of the system, describes the phoneme-based search and provides basic performance metrics for both the English and Czech data.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

20205 - Automation and control systems

Projekt

<a href="/cs/project/TE01020197" target="_blank" >TE01020197: Centrum aplikované kybernetiky 3</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2017

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů