Live TV subtitling through respeaking with remote cutting-edge technology

Pražák, Aleš; Loose, Zdeněk; Psutka, Josef; Radová, Vlasta

doi:10.1007/s11042-019-08235-3

Cited by 7 publications

(6 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other previous studies have recorded good accuracy in caption generation with respeaking techniques. One of the studies recorded an average NER of 98.38%, with a latency of 4.7 s [9]. The most complete study of closed caption quality thus far was conducted with the collaboration of Ofcom in the U.K.…”

Section: Caption Quality Measurementmentioning

confidence: 99%

“…Many studies have focused on the development and analysis of captioning systems for live television programs. Most of these focus on the respeaking method [8][9][10] or ASR methods that mix the respeaking and the original audio as input to the system [11]. These respeaking methods are used to prevent possible problems that may occur in the original audio (noise, mispronunciation of words, speaker changes, etc.).…”

Section: Introductionmentioning

confidence: 99%

“…Using respeaking methods often implies a reduction of the original content (keeping as much of the context of the original content as possible). Therefore, these captions may not cover entire sentences or a speaker's original ideas since parts of the speech (the part considered less relevant) is often discarded, especially to ensure the correct latency of the captions [9]. However, if editing is truly necessary, this must be done with great care as edits are not well received by deaf or hard-ofhearing users who can read lips, and it can be frustrating for them if the captions do not show exactly what the speaker is saying.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Comparative analysis between a respeaking captioning system and a captioning system without human intervention

Ruiz-Arroyo

García-Crespo

Fuenmayor-Gonzalez

et al. 2022

Univ Access Inf Soc

View full text Add to dashboard Cite

People living with deafness or hearing impairment have limited access to information broadcast live on television. Live closed captioning is a currently active area of study; to our knowledge, there is no system developed thus far that produces high-quality captioning results without using scripts or human interaction. This paper presents a comparative analysis of the quality of captions generated for four Spanish news programs by two captioning systems: a semiautomatic system based on respeaking (system currently used by a Spanish TV station) and an automatic system without human interaction proposed and developed by the authors. The analysis is conducted by measuring and comparing the accuracy, latency and speed of the captions generated by both captioning systems. The captions generated by the system presented higher quality considering the accuracy in terms of Word Error Rate (WER between 3.76 and 7.29%) and latency of the captions (approximately 4 s) at an acceptable speed to access the information. We contribute a first study focused on the development and analysis of an automatic captioning system without human intervention with promising quality results. These results reinforce the importance of continuing to study these automatic systems.

show abstract

Section: Caption Quality Measurementmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Comparative analysis between a respeaking captioning system and a captioning system without human intervention

Ruiz-Arroyo

García-Crespo

Fuenmayor-Gonzalez

et al. 2022

Univ Access Inf Soc

View full text Add to dashboard Cite

show abstract

“…Sign language to speech and speech to sign language with a 3D animated avatar along with the subtitles have been proposed [2]. State-of-the-art speech technology for live subtitle system has been designed for public service to be broadcasted in the Czech Republic Television [89]. Bangla speech to sign language translator has been developed using CMU Sphinx tool and macOS.…”

Section: Related Workmentioning

confidence: 99%

An automatic machine translation system for multi-lingual speech to Indian sign language

Dhanjal

Singh

2021

Multimed Tools Appl

View full text Add to dashboard Cite

Sign language (SL) is the best suited communication medium for hearing impaired people. Even with the advancement of technology, there is a communication gap between the hearing impaired and hearing people. The aim of this research work is to bridge this gap by developing an automatic system that translates the speech to Indian Sign Language using Avatar (SISLA). The whole system works in three phases: (i) The first phase includes the speech recognition (SR) of isolated words for English, Hindi and Punjabi in speaker independent environment (ii) The second phase translates the source language into Indian Sign Language (ISL) (iii) HamNoSys based 3D avatar represents the ISL gestures. The four major implementation modules for SISLA include: requirement analysis, data collection, technical development and evaluation. The multi-lingual feature makes the system more efficient. The training and testing speech sample files for English (12,660, 4218), Hindi (12,610, 4211) and Punjabi (12,600, 4193) have been used to train and test the SR models. Empirical results of automatic machine translation show that the proposed trained models have achieved the minimum accuracy of 91%, 89% and 89% for English, Punjabi and Hindi respectively. Sign language experts have also been used to evaluate the sign error rate through feedback. Future directions to enhance the proposed system using non-manual SL features along with the sentence level translation has been suggested. Usability testing based on survey results confirm that the proposed SISLA system is suitable for education as well as communication purpose for hearing impaired people.

show abstract

“…A high speed (more cps) will make it difficult to read the subtitles. It is known that subtitles' speed is partly related to the type of programme being broadcast: for example, in sports programmes, speakers speak very fast in order to report what is happening in the event (Pražák, Psutka, Psutka, & Loose, 2013). In order to determine whether the speed at which the subtitles are being broadcast is adequate, it is necessary to process all the information associated with the broadcast of the subtitles.…”

Section: Introductionmentioning

confidence: 99%

A new system for automatic analysis and quality adjustment in audiovisual subtitled‐based contents by means of genetic algorithms

et al. 2020

View full text Add to dashboard Cite

In Spain, the subtitling service on television for the deaf has been improving in quantity since the General Law on Audiovisual Communication was enacted in 2010. This law establishes a series of quality standards that must be followed in the subtitling process. One of the most relevant aspects of subtitle quality is the speed at which they are shown on the screen, due to the fact that a too high speed (less time on screen) will make them difficult to read and the information hard to understand. In order to determine whether the speed at which the subtitles are being shown is adequate, first, it is necessary to process all the information associated with the broadcast of the digital TV channels including data from different sources. In this research, the authors have worked with the data obtained within the time period between July 2012 and December 2017, that is, with more than 950 million records. This article presents a framework for integration and processing of heterogeneous information associated with the subtitling of audiovisual content from different sources. Moreover, the framework will provide an automatic adjustment of subtitles in broadcasting regarding quality indicators by means of a genetic algorithm approach. The results show that the system is able to estimate the best relationship between the time and size of the subtitles and maintaining the quality levels established for this research. These results have been validated by experts and users of this domain.

show abstract

Live TV subtitling through respeaking with remote cutting-edge technology

Cited by 7 publications

References 14 publications

Comparative analysis between a respeaking captioning system and a captioning system without human intervention

Comparative analysis between a respeaking captioning system and a captioning system without human intervention

An automatic machine translation system for multi-lingual speech to Indian sign language

A new system for automatic analysis and quality adjustment in audiovisual subtitled‐based contents by means of genetic algorithms

Contact Info

Product

Resources

About