1
00:00:00,001 --> 00:00:02,620
[ELECTRONIC MUSIC]

2
00:00:11,360 --> 00:00:15,334
Live Text Access: Training
for real-time interlingual subtitlers

3
00:00:16,800 --> 00:00:18,831
Unit 2. Linguistic competence

4
00:00:18,831 --> 00:00:22,960
Element 1. Functionality:
accuracy,readability and legibility.

5
00:00:24,480 --> 00:00:29,257
What is accuracy?
Created by UAB, SSML and SDI

6
00:00:31,078 --> 00:00:33,769
The Learning outcome
for this video lecture

7
00:00:33,769 --> 00:00:35,753
is to produce accurate transcriptions

8
00:00:35,753 --> 00:00:38,305
in terms of spelling,
grammar and meaning.

9
00:00:38,305 --> 00:00:42,939
In this video lecture
we will be dealing with the following points

10
00:00:42,939 --> 00:00:46,425
Accuracy definition,
Real-time subtitling techniques,

11
00:00:46,425 --> 00:00:50,678
Verbatim versus sensation subtitles 
and Main limitations

12
00:00:50,678 --> 00:00:52,240
we will finish with a Summary.

13
00:00:55,280 --> 00:00:56,736
Accuracy

14
00:00:58,706 --> 00:01:01,977
Accuracy we will first 
provide the definition of accuracy

15
00:01:01,977 --> 00:01:06,433
according to Carlo Eugeni
real-time subtitling can be defined

16
00:01:06,527 --> 00:01:10,240
as the degree to which the subtitle confirms
to a defined standard

17
00:01:10,240 --> 00:01:13,083
Real-time subtitling techniques

18
00:01:14,303 --> 00:01:16,990
Before entering into accuracy
in more detail

19
00:01:16,990 --> 00:01:20,105
it's important to understand
how real-time subtitlers

20
00:01:20,105 --> 00:01:21,269
develop their job.

21
00:01:21,719 --> 00:01:25,552
There are two main techniques
in real-time intralingual subtitling

22
00:01:25,552 --> 00:01:26,735
for the text entry

23
00:01:26,735 --> 00:01:29,651
one is based
on speech recognition systems

24
00:01:29,651 --> 00:01:32,080
and the second is based
on fast typing systems

25
00:01:34,480 --> 00:01:36,451
Speech recognition systems

26
00:01:36,451 --> 00:01:39,526
Professionals working
with speech recognition systems

27
00:01:39,526 --> 00:01:42,800
wear a headset and a microphone 
they are called respeakers.

28
00:01:43,360 --> 00:01:46,814
They listen to speakers
and simultaneously repeat

29
00:01:46,814 --> 00:01:51,295
in word by word or reformulate
what they say to a microphone

30
00:01:51,295 --> 00:01:53,288
connected to the software program

31
00:01:53,288 --> 00:01:56,421
which automatically recognises
the respeaker's voice

32
00:01:56,421 --> 00:01:59,920
and turns it into written 
words that appear on the screen.

33
00:02:02,400 --> 00:02:04,354
Fast typing systems

34
00:02:04,354 --> 00:02:07,431
Professionals working
with fast typing systems

35
00:02:07,431 --> 00:02:10,780
sometimes wear a headset
depending on the setting

36
00:02:10,780 --> 00:02:14,505
and they always work
on special keyboards

37
00:02:14,505 --> 00:02:19,332
they are called stenographers,
plantypists or velotypists to name some

38
00:02:19,332 --> 00:02:22,960
in all cases they listen to the speakers

39
00:02:22,960 --> 00:02:25,723
and simultaneously type word by word

40
00:02:25,723 --> 00:02:28,137
or reformulate what they say
to a special keyboard

41
00:02:28,137 --> 00:02:32,828
based on a phonetic or syllabic system
connected to a software program

42
00:02:33,000 --> 00:02:37,311
which automatically turns
the type test into written words

43
00:02:37,311 --> 00:02:38,661
that appear on the screen.

44
00:02:39,505 --> 00:02:42,567
Verbatim versus Sensatim subtitles

45
00:02:43,562 --> 00:02:47,574
There are two main methods
to subtitle a speech in real time

46
00:02:47,574 --> 00:02:49,917
verbatim and sensatim

47
00:02:49,917 --> 00:02:54,671
verbatim subtitles are required
in some countries such as US or UK.

48
00:02:54,671 --> 00:02:58,480
Accuracy in verbatim means
transcribing every single word

49
00:02:59,120 --> 00:03:02,684
evident mistakes
or features of orality included.

50
00:03:02,684 --> 00:03:05,840
Subtitlers deliver word by word transcriptions

51
00:03:06,480 --> 00:03:10,752
verbatim subtitles
are mostly displayed in rollup mode.

52
00:03:10,752 --> 00:03:15,508
The subtitle appears word by word
at the bottom of the screen

53
00:03:15,508 --> 00:03:20,436
once the bottom line is full
it's pushed upwards by new words.

54
00:03:20,436 --> 00:03:26,090
On the other side sensatim subtitles 
also known as edited subtitles

55
00:03:26,090 --> 00:03:29,365
are required in other countries
mainly European ones

56
00:03:29,365 --> 00:03:32,338
such as Italy, Spain or Germany.

57
00:03:32,338 --> 00:03:35,966
In this case the subtitler
is more focused on the meaning

58
00:03:35,966 --> 00:03:38,489
and the original speech of the speaker
is edited

59
00:03:38,489 --> 00:03:42,000
by omitting unnecessary utterances

60
00:03:42,000 --> 00:03:44,551
or summarising quick utterances

61
00:03:44,551 --> 00:03:50,327
to allow the text version of the speech
to flow more easily for the reader

62
00:03:50,684 --> 00:03:55,779
Sensatim or edited subtitles
are mostly displayed in pop-up modes

63
00:03:55,779 --> 00:04:00,160
in which a block subtitle appears
and disappears at once

64
00:04:02,320 --> 00:04:03,520
Main limitations

65
00:04:07,040 --> 00:04:08,342
Not enough time!

66
00:04:08,342 --> 00:04:12,560
It is important to bear in mind
that there is not always enough time.

67
00:04:13,520 --> 00:04:17,916
Most times real-time subtitlers
fight against the clock

68
00:04:17,916 --> 00:04:21,362
time constraints will depend
on the working context

69
00:04:21,362 --> 00:04:24,065
in which professionals
carry out their jobs.

70
00:04:24,065 --> 00:04:27,644
For instance the news
and the parliamentary sessions

71
00:04:27,644 --> 00:04:29,656
are particularly fast

72
00:04:29,656 --> 00:04:33,168
and professionals 
have to go along with speakers

73
00:04:33,168 --> 00:04:36,300
because form
is as important as content

74
00:04:36,300 --> 00:04:41,405
on the other side real-time subtitlers
working with sports programs

75
00:04:41,405 --> 00:04:44,668
have to describe actions
shown on the screen

76
00:04:44,668 --> 00:04:47,130
rather than speech utterances

77
00:04:47,130 --> 00:04:50,837
so there will be much more edit
on sports

78
00:04:50,837 --> 00:04:57,520
than with news or the parliament
regardless of how fast the speaker is

79
00:04:59,600 --> 00:05:02,002
Limitations
as Lambourne states:

80
00:05:02,002 --> 00:05:06,770
" all real-time intralingual
subtitling techniques share the same limitations"

81
00:05:06,770 --> 00:05:09,435
creating and delivering subtitles
in real-time

82
00:05:09,435 --> 00:05:13,542
for a truly live unscribed TV program,
meeting or live event

83
00:05:13,542 --> 00:05:15,997
involves a degree of compromise

84
00:05:15,997 --> 00:05:19,004
perfection is not achievable
almost by definition

85
00:05:19,004 --> 00:05:22,764
because there is simply
not enough time in which

86
00:05:22,764 --> 00:05:27,684
to carefully edit the text transcribe it
and correct any errors

87
00:05:27,684 --> 00:05:31,673
time it to match
what the speaker is saying

88
00:05:31,673 --> 00:05:35,326
position it
so as never to obscure

89
00:05:35,326 --> 00:05:37,392
interesting on-screen information

90
00:05:37,760 --> 00:05:42,240
presented in a way that the viewer
has always has time to read it.

91
00:05:45,600 --> 00:05:50,068
Aim real-time subtitlers should be able

92
00:05:50,068 --> 00:05:54,480
to reproduce the source text
without losing key information

93
00:05:56,640 --> 00:05:57,842
Summary

94
00:05:59,118 --> 00:06:02,843
As a summary we would like
to stress the following points

95
00:06:02,843 --> 00:06:08,209
First, there are two main techniques
in real-time subtitling

96
00:06:08,209 --> 00:06:11,840
voice recognition systems
and fast typing systems.

97
00:06:11,840 --> 00:06:15,745
Second, accuracy will depend
on the subtitling method

98
00:06:15,745 --> 00:06:17,656
verbatim or sensatim.

99
00:06:17,656 --> 00:06:22,160
Third, time constraints 
are always a limitation for accuracy.

100
00:06:23,280 --> 00:06:24,537
Exercises

101
00:06:25,607 --> 00:06:29,607
The exercises for this video lecture 
are in the Trainer's Guide

102
00:06:29,607 --> 00:06:31,269
and the PowerPoint file.

103
00:06:31,926 --> 00:06:34,489
[ELECTRONIC MUSIC]