1
00:00:07,450 --> 00:00:10,780
EASIT: Easy access
for social inclusion training.
2
00:00:12,600 --> 00:00:17,100
This is Unit 3B: Easy to understand
and audio description.
3
00:00:17,100 --> 00:00:19,620
Element 1: Processes.
4
00:00:19,850 --> 00:00:24,180
Video lecture: What are
easy to understand audio subtitles.
5
00:00:24,650 --> 00:00:28,800
I am Elisa Perego,
from Università degli Studi di Trieste.
6
00:00:28,800 --> 00:00:33,100
In this video lecture
I will define audio subtitles,
7
00:00:35,260 --> 00:00:39,060
and suggest ways to simplify them.
8
00:00:40,100 --> 00:00:45,420
Think of the German-American
miniseries Unorthodox,
9
00:00:45,420 --> 00:00:51,100
about a girl rejecting
her radicalised upbringing.
10
00:00:51,550 --> 00:00:55,380
In the series,
there is a constant code-switching
11
00:00:55,390 --> 00:00:57,740
from Yiddish, to English, to German.
12
00:00:58,900 --> 00:01:00,620
In the original version,
13
00:01:00,630 --> 00:01:04,660
Yiddish and German
are subtitled in English.
14
00:01:04,980 --> 00:01:07,870
But how can subtitles be accessed
15
00:01:07,870 --> 00:01:10,380
by a person with sight loss or dyslexia,
16
00:01:10,380 --> 00:01:11,860
by a slow reader,
17
00:01:11,870 --> 00:01:15,220
by a person far away from the screen
18
00:01:15,230 --> 00:01:20,460
or by anyone doing something
in another room?
19
00:01:20,800 --> 00:01:24,420
There is only one solution:
conveying subtitles orally.
20
00:01:24,430 --> 00:01:27,800
Which means reading them aloud
21
00:01:27,810 --> 00:01:31,780
and integrating them with AD.
22
00:01:31,850 --> 00:01:37,800
Audio subtitles are as short
and focused as subtitles.
23
00:01:37,800 --> 00:01:43,140
But this does not mean
they are easy to understand.
24
00:01:43,780 --> 00:01:48,700
How can audio subtitles
be more understandable?
25
00:01:49,000 --> 00:01:54,020
We can briefly introduce them
26
00:01:54,020 --> 00:01:59,180
to offer background to the user
and prevent confusion.
27
00:02:00,000 --> 00:02:03,180
A short text
in easy-to-understand language
28
00:02:03,540 --> 00:02:07,580
written on screen and read aloud can do.
29
00:02:08,150 --> 00:02:12,100
For example:
"This film includes audio subtitles".
30
00:02:12,100 --> 00:02:16,740
"Audio subtitles
are subtitles read aloud".
31
00:02:16,900 --> 00:02:19,260
In fully audio subtitled films,
32
00:02:19,260 --> 00:02:24,100
using multiple, acted voices
for each character
33
00:02:24,300 --> 00:02:30,300
helps us better enjoy, understand
and remember the film.
34
00:02:30,900 --> 00:02:35,780
Human-voiced audio subtitles
should be preferred
35
00:02:35,780 --> 00:02:38,260
to text-to-speech audio subtitles.
36
00:02:38,260 --> 00:02:41,460
In fact, texto-to-speech audio subtitles
37
00:02:41,460 --> 00:02:45,980
can interfere negatively
with the listening experience.
38
00:02:46,900 --> 00:02:50,280
Another solution is adapting
the subtitle formulation
39
00:02:50,290 --> 00:02:53,560
if the subtitles are too complex.
40
00:02:53,600 --> 00:02:57,810
To do so, you can merge subtitles
for the deaf and hard of hearing
41
00:02:57,810 --> 00:03:00,000
with easy-to-understand principles.
42
00:03:00,600 --> 00:03:04,740
Let's take the subtitle
from the film I, Tonya.
43
00:03:04,990 --> 00:03:09,180
I am so sorry,
but there is no smoking on the ice.
44
00:03:09,200 --> 00:03:13,700
It can become
a third-person audio subtitle:
45
00:03:13,700 --> 00:03:19,540
The teacher kindly asks LaVona
not to smoke on the ice rink.
46
00:03:19,800 --> 00:03:25,780
The apology I am so sorry
is replaced by an adverb: kindly.
47
00:03:26,200 --> 00:03:31,940
The adjunct on the ice, whose meaning
is clearer if the visuals are accessed,
48
00:03:32,200 --> 00:03:37,340
is replaced by an expression
that refers explicitly
49
00:03:37,340 --> 00:03:41,900
to the area of ice for skating:
the ice rink.
50
00:03:42,800 --> 00:03:45,100
If ice rink is too jergal,
51
00:03:45,100 --> 00:03:49,260
ice-skating spot
can be a simpler choice.
52
00:03:50,060 --> 00:03:53,300
If the audio subtitle is too long
53
00:03:53,300 --> 00:03:56,300
the adverb kindly can be removed,
54
00:03:56,300 --> 00:03:59,620
without major content consequences.
55
00:04:00,700 --> 00:04:06,220
Easy-to-understand
audio subtitlers can consider
56
00:04:06,220 --> 00:04:08,780
the main issues of struggling users,
57
00:04:08,900 --> 00:04:13,980
like their poor lexical
and structural knowledge.
58
00:04:14,000 --> 00:04:18,140
A solution is choosing mainly
high frequency words,
59
00:04:18,140 --> 00:04:20,820
short declarative sentences,
60
00:04:20,820 --> 00:04:24,260
or normalising idiomatic formulations.
61
00:04:25,540 --> 00:04:27,580
Back to the smoking example,
62
00:04:27,600 --> 00:04:31,260
direct speech can be used.
63
00:04:31,270 --> 00:04:35,020
Teacher: It is forbidden to smoke here,
64
00:04:35,100 --> 00:04:39,020
with a word reduction from 11 to 7.
65
00:04:39,900 --> 00:04:45,900
We know that reduction affects
the interpersonal function of language.
66
00:04:46,250 --> 00:04:48,980
Restoring it through intonation,
67
00:04:48,980 --> 00:04:51,940
a kind mellow one in this case,
68
00:04:51,940 --> 00:04:53,620
is crucial.
69
00:04:53,700 --> 00:04:57,300
In general, avoiding a flat style
70
00:04:57,300 --> 00:05:01,940
and using all the prosodic possibilities
of the human voice
71
00:05:02,100 --> 00:05:05,660
can ensure that E2U audio subtitles
72
00:05:05,660 --> 00:05:09,340
really catch
the attention of the listener.
73
00:05:10,140 --> 00:05:11,940
On a more technical level,
74
00:05:11,970 --> 00:05:14,740
we could consider allowing the user
75
00:05:14,740 --> 00:05:17,940
to manage the volume
of the audio subtitles
76
00:05:17,940 --> 00:05:21,620
separately from the volume
of the original sound.
77
00:05:21,980 --> 00:05:24,820
Many more solutions exist.
78
00:05:25,000 --> 00:05:29,860
The topic is wide,
there are many factors to explore,
79
00:05:29,860 --> 00:05:32,500
and many questions to answer.
80
00:05:32,600 --> 00:05:34,480
I will conclude with a couple:
81
00:05:34,500 --> 00:05:37,500
Would automatic audio subtitling work
82
00:05:37,500 --> 00:05:40,140
in an easy-to-understand context?
83
00:05:40,200 --> 00:05:44,460
And, can we really consider
using synthetic voices
84
00:05:44,460 --> 00:05:49,800
in contexts where involvement
and clarity are a priority?
85
00:05:50,700 --> 00:05:53,500
All the pictures used
in this presentation
86
00:05:53,600 --> 00:05:56,620
come from
the Public Domain Vectors website,
87
00:05:56,620 --> 00:06:00,580
which gathers hundreds
or royalty-free vector images
88
00:06:00,580 --> 00:06:06,060
under the Creative Commons Universal
Public Domain Dedication licence.
89
00:06:07,500 --> 00:06:11,040
This video lecture was prepared
by Elisa Perego
90
00:06:11,100 --> 00:06:14,420
and produced by Andreea Deleanu
91
00:06:14,420 --> 00:06:17,100
from Università degli Studi di Trieste,
92
00:06:17,200 --> 00:06:21,060
with the help of Ester Hedberg
and Anna Matamala.
93
00:06:21,700 --> 00:06:27,620
You can reach me at eperego@units.it.