Top Banner
SURFnet Relatiedagen 2010 Noordwijkerhout December 9, 2010 Searching in spoken words Disclosure of recorded content in MediaMosa Speech & Language Technology [email protected]
24

Transcription verhaal2010

Jul 07, 2015

Download

Documents

MediaMosa

Searching in spoken words . Disclosure of recorded content in MediaMosa.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Transcription verhaal2010

SURFnet Relatiedagen 2010Noordwijkerhout December 9, 2010

Searching in spoken wordsDisclosure of recorded content in MediaMosa

Speech & Language

[email protected]

Page 2: Transcription verhaal2010

• Introduction

– Why speech is so important

– What is HLT?

• Working applications:

– Self-service (Internet & Telephony)

– Searching in recorded audiovisual recordings

• Demonstrations

Page 3: Transcription verhaal2010

What is HLT?

• Human Language Technology is the technology that mimics the human language capacity.

speech text sign

Page 4: Transcription verhaal2010

Redundancy

• Vlgones een oznrdeeok op een Eglneseuvinretsiet mkaat het neit uitin wlkeevloogdre de ltteers in een wrood saatn, het einge watblegnaijrk is is dat de eretse en de ltaatse ltteer op de jiutsepatals saatn. De rset van de ltteers mgoen wllikueirggpletaastwdoren en je knut vrelvogensgwoeon lzeen wat er saatt. Dit kmotodmatwe neit ekle ltteer op zcih lzeen maar het wrood als gheeel.

Page 5: Transcription verhaal2010
Page 6: Transcription verhaal2010

WORKING APPLICATIONS

Dialogue systems (telephony, real time, limited complexity)

Disclosure systems (high quality audio, offline, complex)

Page 7: Transcription verhaal2010

Dictation

Voice

Information Retrieval

Mens-Machine-Communicatie

Emotie detectie:Lachen/Huilen

Spoken Document Retrieval

Web

Mobile

ContactCenter

Natural Language Search

HLT

Page 8: Transcription verhaal2010

Companies using speech technology

Page 9: Transcription verhaal2010

How may I help you

Who is calling?Identification via ZIP-

code and house number

Why are they

calling?Classification based on

the recognition of the

question: “how may I

help you”

Page 10: Transcription verhaal2010

Organisations using speech technology

Page 11: Transcription verhaal2010

Disclosure of audiovisual archives

• The number of AV-archives on the Internet increases rapidly

• Archiving is not enough: disclosure and reusing is required!

• The use of HLT is needed (humans cost too much).

Page 12: Transcription verhaal2010

Buchenwald

Digitalized (historic)

collections

H.M. Koningin

Wilhelmina

Digital recorded collections

Second feministic wave

Memories of Indonesia

WFH

LVSR

Page 13: Transcription verhaal2010

Searching in historic radio recordings:Radio Oranje

Page 14: Transcription verhaal2010

Oral History: Buchenwald

Page 15: Transcription verhaal2010

Oral History: Brandgrens, Rotterdam

10 getuigen van het bombardement van Rotterdam (mei ‘40) vertellen hun verhaal. TST wordt gebruikt om in de getuigenissen te zoeken.

Page 16: Transcription verhaal2010

Searching in the radio interviews of WFH

Page 17: Transcription verhaal2010

Searching in 46 interview collections:getuigenverhalen (600 hour)

Page 18: Transcription verhaal2010

Searching in 500 interviews in Croatia

Page 19: Transcription verhaal2010

CroMe - Audio Search

Search word traumas

Language

found

Page 20: Transcription verhaal2010

Political metings

Page 21: Transcription verhaal2010

Parliament

transcriptions

Gisteren was er een bespreking ivm de betrekkingen tussen Nederland en Vlaanderen

Page 22: Transcription verhaal2010

Recognition of lectures

• Record the speech

• Record the PPT

• Recognise the speech

• Use the display time of each slide as THE time unit

• Use the recognised speech as keywords for each slide

Page 23: Transcription verhaal2010

Searching in news broadcasts

Page 24: Transcription verhaal2010

Questions?