International Telecommunication Union

1
ITU Workshop on Making Media Accessible to
allThe options and the economics (Geneva,
Switzerland, 24 (p.m.) 25 October 2013)
Access Services for TVQuality Metrics and
Challenges in the Provision of Live Closed
Captioning in Brazil
Gabriela Campedelli Cidade do Conhecimento USP,
Brazil ltgcampedelli_at_gmail.comgt
Geneva, Switzerland, 25 October 2013
2
Background data

• Population 201 million
• Area 8.5 million km2
• An estimated 16 million Brazilians (8) have
  special needs
• Free-to-air TV still important
• Internet surfing approaching same numbers as TV.

Geneva, Switzerland, 25 October 2013
2
3
TV in Brazil

• Analogue terrestrial broadcasting
• Federative broadcasting model
• All local stations
• Network composed by local content affiliates
• Digital terrestrial TV ISDB-T started in 2007
• Analog swicth-off planned for 2016 (option to
  extend deadline)

Geneva, Switzerland, 25 October 2013
3
4
TV genres in Brazil

• The programming provider station and affiliates
  all offer pre-recorded and live programming
• News and current affairs
• General entertainment
• Telenovelas
• Pre-recorded soaps and series
• Movies
• Reality shows
• Music shows

Geneva, Switzerland, 25 October 2013
4
5
TV access services in Brazil

• Demand
• 5 million individuals (2.5) who are deaf or
  have recognized, serious hearing impairments
• Supply
• 14 hours daily (from 6 am to 2 am) regulator
  only counts 2 hours from 2 am to 6 am
• 6 hours (6-14) and 6 hours (17 02)
• Regulatory framework applies to all
  free-to-air TV in cities with 500,000
  inhabitants

Geneva, Switzerland, 25 October 2013
5
6
TV captioning - metrics

• NORMA BRASILEIRA ABNT NBR 15290
• Primeira edição 31.10.2005 Acessibilidade em
  comunicação na televisão Accessibility in TV
  captions
• Predates legislation and audience research on
  deafness and accessibility!
• Based on old US and Canadian standards
• Specifications for captioning, video description
  and sign language

Geneva, Switzerland, 25 October 2013
6
7
TV captioning - metrics

• Closed captioning
• Captioning delay max. 4 seconds (live) or
  frame-accurate (pre-recorded content) c.f. 4.1.9
  Sincronia
• Minimum/maximum display times depending on
  whether there are one, two or three lines of
  captioning c.f. 4.2.8 Tempo de exposição
• Verbatim captioning for live closed captioning
• 98 accuracy threshold for live captioning

Geneva, Switzerland, 25 October 2013
7
8
TV captioning - metrics

• No specification of how this is to be measured
  (c.f. NER model Spain AENOR 15390)
• NCRA in USA uses as parameter for quality of
  transcriptions for live stenotype but these are
  not mandatory there
• No mention of the research on which these metrics
  were based

Geneva, Switzerland, 25 October 2013
8
9
TV captioning - evidence

• Users in Brazil report inobservancy in ABNT
  metrics
• Synchronicity in ABNT metrics are impossible to
  achieve
• BBCs recent paper by Mike Armstrong
• demonstrated the link between syncronicity and
  perceived quality by viewers.
• Source Mike Amstrong (2013). THE
  DEVELOPMENT OF A METHODOLOGY TO EVALUATE THE
  PERCEIVED QUALITY OF LIVE TV SUBTITLES. BBC RD,
  UK, IBC 2013.

Geneva, Switzerland, 25 October 2013
9
10
TV captioning - evidence
Geneva, Switzerland, 25 October 2013
10
11
TV captioning - evidence

• "The clearest trend for timing was for people
  watching with sound where there was a strong and
  statistically significant increase in the quality
  score with improved timing (reduced delay).
• For the range of timings tested, each 1 second
  reduction in the subtitle delay gave just over 5
  points improvement in the quality score."

Geneva, Switzerland, 25 October 2013
11
12
TV captioning - evidence

• The presentation time of each caption also
  depends on
• Presentation form (Pop-up or Roll-on)
• Liguistic issues (verbatim or linguistic
  condensation to reduce the required reading
  speed).
• Are we using the research that has already be
  done in this field?

Geneva, Switzerland, 25 October 2013
12
13
TV captioning other issues affecting quality

• Analog TV
• Atmospheric interference (heavy rain etc)
• Strange characters due to problems in signal
  transmission over country
• Telecommunications problems
• Digital TV
• Blocking and other artifacts.

Geneva, Switzerland, 25 October 2013
13
14
TV captioning options to improve quality

• Reducing the delay in live CC
• Reviewing the service end-to-end to identify ways
  of reducing the delay
• Issues with Emergency Alerts
• Improvements in
• Re-speaking (speech recognition)
• Stenography (adjustments in brazilian portuguese
  vocabulary)

Geneva, Switzerland, 25 October 2013
14
15
Other Technologies in Accessibility

• LIBRAS APP signing language
• Prize WSA-Mobile
• Possible use in second
• Screen

Geneva, Switzerland, 24 October 2013
15
16
Other uses of closed captioning

• BOX Fish (TV API) search engine (USA)
• Commercial app in second screen market
  http//www.niemanlab.org/2012/12/opened-captions-t
  urning-the-spoken-words-on-tv-screens-into-streams
  -of-hackable-data/
• Could accessibility services in communication be
  called metadata in order to include technology in
  market?

Geneva, Switzerland, 24 October 2013
16
17
TV captioning conclusions

• Setting tough mandatory quality metrics is no
  guarantee of captioning quality if the targets
  are impossible to achieve
• Quality metrics can vary from language to
  language
• Cultural factors are involved
• Technology methodology crucial
• Need for more research with audience to deliver
  closed captioning to promote understanding
• Understanding is the best metric

Geneva, Switzerland, 25 October 2013
17
18
Thank you!

• Gabriela Campedelli
• 55 11 97637 9099
• gcampedelli_at_gmail.com
• http//www.cidade.usp.br
• LinkedIn http//www.linkedin.com/profile/view?id
  203273681trkwvmp-profile
• Lattes http//buscatextual.cnpq.br/buscatextual/v
  isualizacv.do?idK4228482E6

Geneva, Switzerland, 25 October 2013
18