Automatic Live Subtitling: state of the art, expectations and current trends
Authors: Carlo Aliprandi, Cristina Scudellari, Isabella Gallucci, Nicola Piccinini, Matteo Raffaelli, Arantza del Pozo, Aitor Álvarez, Haritz Arzelus, Renato Cassaca, Tiago Luis, Joao Neto, Carlos Mendes, Sérgio Paulo, Marcio Viveiros
Date: 07.04.2014
Abstract
The subtitling demand has grown quickly over the years. The path of manual subtitling is no longer feasible, due to increased costs and reduced production times. Assisted Subtitling is an emerging technique, consisting in the application of Automatic Speech Recognition (ASR) to automatically generate program transcripts. This paper will report on recent advances in ASR, presenting SAVAS, a novel Speaker Independent ASR technology specifically designed for Live Subtitling. We will describe the technology, presenting its features and detailing language and domain-specific tunings that we have carried out. We will also introduce the S.Scribe!, S.Live! and S.Respeak! systems, which are based on SAVAS. S.Scribe! is a batch Speaker Independent Transcription system for subtitling. S.Live! is a first-of-a-kind Speaker Independent Transcription System, with real-time performances for online subtitling. S.Respeak! is a collaborative Respeaking System, for live and batch production of multilingual subtitles. S.Respeak! has proven to be sufficiently robust forprograms where the acoustic conditions are challenging and for spontaneous speech. Similar results are expected tobe achieved also for S.Live! and S.Scribe!, which are currently being tested under real conditions at different broadcasters premises, to subtitle live programs, in both assisted and unassisted tasks. We will finally detail performances of the systems for 7 languages (English, Spanish, Italian,French, German, Portuguese and Basque).
BIB_text
author = {Carlo Aliprandi, Cristina Scudellari, Isabella Gallucci, Nicola Piccinini, Matteo Raffaelli, Arantza del Pozo, Aitor Álvarez, Haritz Arzelus, Renato Cassaca, Tiago Luis, Joao Neto, Carlos Mendes, Sérgio Paulo, Marcio Viveiros},
title = {Automatic Live Subtitling: state of the art, expectations and current trends },
pages = {141-148},
keywds = {
automatic subtitling, speech recognition
}
abstract = {
The subtitling demand has grown quickly over the years. The path of manual subtitling is no longer feasible, due to increased costs and reduced production times. Assisted Subtitling is an emerging technique, consisting in the application of Automatic Speech Recognition (ASR) to automatically generate program transcripts. This paper will report on recent advances in ASR, presenting SAVAS, a novel Speaker Independent ASR technology specifically designed for Live Subtitling. We will describe the technology, presenting its features and detailing language and domain-specific tunings that we have carried out. We will also introduce the S.Scribe!, S.Live! and S.Respeak! systems, which are based on SAVAS. S.Scribe! is a batch Speaker Independent Transcription system for subtitling. S.Live! is a first-of-a-kind Speaker Independent Transcription System, with real-time performances for online subtitling. S.Respeak! is a collaborative Respeaking System, for live and batch production of multilingual subtitles. S.Respeak! has proven to be sufficiently robust forprograms where the acoustic conditions are challenging and for spontaneous speech. Similar results are expected tobe achieved also for S.Live! and S.Scribe!, which are currently being tested under real conditions at different broadcasters premises, to subtitle live programs, in both assisted and unassisted tasks. We will finally detail performances of the systems for 7 languages (English, Spanish, Italian,French, German, Portuguese and Basque).
}
date = {2014-04-07},
year = {2014},
}
