1
00:00:06,080 --> 00:00:08,760
Interpreter: If you create
video and audio content

2
00:00:08,760 --> 00:00:10,120
or host events,

3
00:00:10,120 --> 00:00:12,520
there are two reasons
to care about providing

4
00:00:12,520 --> 00:00:16,320
high quality, same language
captioning and transcription.

5
00:00:16,320 --> 00:00:18,260
The first is obvious,

6
00:00:18,260 --> 00:00:20,700
that there are millions
of people like myself,

7
00:00:20,700 --> 00:00:22,940
who get accessibility through captions.

8
00:00:22,940 --> 00:00:24,130
But another reason


9
00:00:24,130 --> 00:00:27,210
is that you will reach
a much larger audience

10
00:00:27,210 --> 00:00:29,130
and increase your return
on investment.

11
00:00:29,130 --> 00:00:33,280
80% of people who use
captions are not deaf.

12
00:00:33,280 --> 00:00:36,100
Captions help foreign language learners,

13
00:00:36,100 --> 00:00:40,430
people in gyms, or bars, 
or sound sensitive environments,

14
00:00:40,430 --> 00:00:43,750
and assist when the speaker
has a strong accent,

15
00:00:43,750 --> 00:00:48,140
or when content is complicated
and difficult to understand.

16
00:00:48,140 --> 00:00:51,690
Captions increase
the viewership by 40%.

17
00:00:51,690 --> 00:00:55,210
90% of videos with captions
are watched to completion.

18
00:00:55,210 --> 00:00:56,390
Digiday states that

19
00:00:56,390 --> 00:00:59,980
85% of Facebook videos
are viewed in mute mode,

20
00:00:59,980 --> 00:01:02,700
so captioning has become a necessity.

21
00:01:02,700 --> 00:01:05,040
Growing up,
I didn't have captioning access

22
00:01:05,040 --> 00:01:07,350
until I was in eighth grade.

23
00:01:07,350 --> 00:01:09,370
That's when my dad brought home a box

24
00:01:09,370 --> 00:01:11,400
called a captioning decoder.

25
00:01:11,400 --> 00:01:14,070
We were amazed by this magic box

26
00:01:14,070 --> 00:01:17,260
that made captions show up
on almost all the TV channels.

27
00:01:17,260 --> 00:01:21,010
Now captioning support
is included in all TV models

28
00:01:21,030 --> 00:01:22,600
and video players,

29
00:01:22,600 --> 00:01:25,420
and there's no need 
for external decoders anymore.

30
00:01:25,420 --> 00:01:29,250
Captions are now not limited
even to TV and videos,

31
00:01:29,250 --> 00:01:31,770
they can also be provided
in real-time

32
00:01:31,770 --> 00:01:34,780
for breaking news, webinars, events,

33
00:01:34,780 --> 00:01:37,670
classes, work meetings, and so on.

34
00:01:37,670 --> 00:01:40,650
Once I was given access
to captioned material,

35
00:01:40,650 --> 00:01:42,380
my life completely changed

36
00:01:42,380 --> 00:01:45,000
and new worlds opened up for me.

37
00:01:45,000 --> 00:01:46,170
But I'm not alone.

38
00:01:46,170 --> 00:01:49,470
There are over 48 million 
deaf and hard of hearing people

39
00:01:49,470 --> 00:01:50,720
in United States.

40
00:01:50,720 --> 00:01:54,720
And over 466 million worldwide.

41
00:01:54,720 --> 00:01:57,150
The number is increasing
rapidly as well,

42
00:01:57,150 --> 00:02:00,170
especially because the number
one cause of hearing loss now

43
00:02:00,170 --> 00:02:03,323
is constant exposure to loud noises.

44
00:02:05,160 --> 00:02:09,470
Captions also helped me master
English as my third language.

45
00:02:09,470 --> 00:02:12,330
Even my parents,
who have perfect hearing,

46
00:02:12,330 --> 00:02:15,793
love captions because English
is not their native language.

47
00:02:18,180 --> 00:02:20,425
Quality is critical

48
00:02:20,425 --> 00:02:23,030
when you're considering captions.

49
00:02:23,030 --> 00:02:25,980
While there are now
many speech technologies

50
00:02:25,980 --> 00:02:27,950
that can translate automatically

51
00:02:27,950 --> 00:02:29,450
spoken words into captions,

52
00:02:29,450 --> 00:02:31,530
they're often not accurate.

53
00:02:31,530 --> 00:02:34,503
Many deaf people call them
"craptions".

54
00:02:34,503 --> 00:02:35,336
(Laughter)

55
00:02:35,336 --> 00:02:36,870
Under ideal conditions,

56
00:02:36,870 --> 00:02:40,837
the accuracy rate is usually
somewhere between 80 and 90%.

57
00:02:40,837 --> 00:02:43,890
And it gets even worse
with complex content,

58
00:02:43,890 --> 00:02:46,070
or accents, background noise,

59
00:02:46,070 --> 00:02:48,120
when speakers are speaking too fast,

60
00:02:48,120 --> 00:02:49,970
or overlap each other.

61
00:02:49,970 --> 00:02:52,460
And bad captions 
are not better than nothing.

62
00:02:52,460 --> 00:02:55,240
Garbled captions are hard to read.

63
00:02:55,240 --> 00:02:58,490
And it's just like
listening to garbled audio.

64
00:02:58,490 --> 00:03:00,790
Caption errors make it hard to understand

65
00:03:00,790 --> 00:03:03,250
or retain any information.

66
00:03:03,250 --> 00:03:07,570
Poor captions affect your SEO,
search engine optimization,

67
00:03:07,570 --> 00:03:10,750
and decrease your CPM,
your clicks per mille,

68
00:03:10,750 --> 00:03:14,763
for videos and your ROI,
return on investment.

69
00:03:15,830 --> 00:03:18,490
Google is deaf like me

70
00:03:18,490 --> 00:03:21,100
and doesn't like videos
with auto-captioning,

71
00:03:21,100 --> 00:03:24,140
so it just does not index
those videos.

72
00:03:24,140 --> 00:03:26,200
So Google penalizes producers

73
00:03:26,200 --> 00:03:27,970
for turning on the auto-captions

74
00:03:27,970 --> 00:03:30,770
instead of actually adding
proper captioning themselves.

75
00:03:32,320 --> 00:03:34,610
Even if a machine could translate

76
00:03:34,610 --> 00:03:37,070
every spoken word
to captions accurately,

77
00:03:37,070 --> 00:03:40,310
it still cannot follow
certain quality guidelines.

78
00:03:40,310 --> 00:03:43,380
For example, you may see
text exactly as it's heard,

79
00:03:43,380 --> 00:03:44,873
for example this:

80
00:03:46,090 --> 00:03:47,673
Lets eat grandma.

81
00:03:48,593 --> 00:03:49,670
(Laughter)

82
00:03:49,670 --> 00:03:52,280
Now, does that mean
that we should go ahead

83
00:03:52,280 --> 00:03:53,930
and consume grandma,

84
00:03:53,930 --> 00:03:55,520
or that we should dine with her?

85
00:03:55,520 --> 00:03:57,140
It's not clear, right?

86
00:03:57,140 --> 00:03:58,706
Now that's much better.

87
00:03:58,706 --> 00:04:00,360
(Laughter)

88
00:04:00,360 --> 00:04:03,320
So proper grammar 
and punctuation are critical

89
00:04:03,320 --> 00:04:05,700
for optimal experience
in reading captions.

90
00:04:05,700 --> 00:04:08,450
Just like speech intonations are.

91
00:04:08,450 --> 00:04:10,760
Quality guidelines are long and complex,

92
00:04:10,760 --> 00:04:14,020
so I'm just sharing a few
basic tips with you today.

93
00:04:14,020 --> 00:04:17,019
Transcripts and captions
need to be verbatim

94
00:04:17,019 --> 00:04:19,240
They need to follow
proper grammatical rules,

95
00:04:19,240 --> 00:04:21,410
include non-speech elements,

96
00:04:21,410 --> 00:04:23,660
use upper and lower case letters,

97
00:04:23,660 --> 00:04:26,580
have good font size and color contrast,

98
00:04:26,580 --> 00:04:28,520
and translate profanity --

99
00:04:29,480 --> 00:04:30,330
word for word,

100
00:04:30,330 --> 00:04:32,480
assuming it's not bleeped out 
in the audio.

101
00:04:34,920 --> 00:04:38,690
Captions and transcripts 
need to be completed

102
00:04:38,690 --> 00:04:41,710
before you publish 
or share videos and podcasts.

103
00:04:44,010 --> 00:04:46,810
So everyone can enjoy them
at the same time.

104
00:04:46,810 --> 00:04:48,430
And if you advertise in advance

105
00:04:48,430 --> 00:04:50,640
that events will have real-time captions,

106
00:04:50,640 --> 00:04:52,363
you will get a larger audience.

107
00:04:54,380 --> 00:04:56,180
With advances in technologies,

108
00:04:56,180 --> 00:04:58,430
it is much easier, faster,and cheaper

109
00:04:58,430 --> 00:05:01,690
to create captions now
more than ever before.

110
00:05:01,690 --> 00:05:04,820
For example, 
YouTube offers a great free tool

111
00:05:04,820 --> 00:05:06,980
to create and edit captions.

112
00:05:06,980 --> 00:05:11,080
You can save and download
your captions as an SRT caption file

113
00:05:11,120 --> 00:05:14,530
and import it onto Vimeo, Facebook,

114
00:05:14,530 --> 00:05:16,220
or other video platforms

115
00:05:16,220 --> 00:05:18,630
that offer captioning support.

116
00:05:18,630 --> 00:05:21,240
Here's a short example.

117
00:05:21,240 --> 00:05:23,313
So you can see how easy it is.

118
00:05:25,400 --> 00:05:27,063
Yes, it's really easy.

119
00:05:28,340 --> 00:05:30,740
And you can clean up 
the auto captions if you like,

120
00:05:30,740 --> 00:05:32,500
or you can create them from scratch.

121
00:05:32,500 --> 00:05:34,740
Just please do not turn on
auto captions

122
00:05:34,740 --> 00:05:36,600
and just leave them as is.

123
00:05:36,600 --> 00:05:38,610
And please add video captions

124
00:05:38,610 --> 00:05:40,680
on all platforms that you use,

125
00:05:40,680 --> 00:05:44,720
and couple those captions
with transcripts for videos,

126
00:05:44,720 --> 00:05:48,940
because those are useful 
for people who cannot see the video

127
00:05:48,940 --> 00:05:50,853
or that want to skim the information.

128
00:05:51,820 --> 00:05:53,610
If your organization can do so,

129
00:05:53,610 --> 00:05:56,080
consider hiring an experienced specialist

130
00:05:56,080 --> 00:05:59,650
who is familiar with quality
standards and guidelines.

131
00:05:59,650 --> 00:06:02,510
They can produce your captions
and transcripts faster

132
00:06:02,510 --> 00:06:05,200
and make them look 
more professional and easy to read.

133
00:06:05,200 --> 00:06:08,520
The costs for transcripts and captions

134
00:06:08,520 --> 00:06:12,200
vary from a dollar all the way 
to $10 or $15 per minute.

135
00:06:12,200 --> 00:06:14,790
And the rates really depend

136
00:06:14,790 --> 00:06:17,823
on experience and the skills 
of the vendor,

137
00:06:19,160 --> 00:06:21,290
the audio quality,

138
00:06:21,290 --> 00:06:25,600
the content complexity,
among other factors.

139
00:06:25,600 --> 00:06:29,380
Low cost may result in bad quality.

140
00:06:29,380 --> 00:06:32,450
Some vendors may use
auto-captioning themselves

141
00:06:32,450 --> 00:06:34,660
or unskilled labor

142
00:06:34,660 --> 00:06:36,860
that will result in low caption quality

143
00:06:36,860 --> 00:06:38,810
and end up wasting your money.

144
00:06:38,810 --> 00:06:42,090
Always review 
the selected vendor's prior work

145
00:06:42,090 --> 00:06:43,533
to assess the quality.

146
00:06:44,660 --> 00:06:47,660
If you have budget limitations,

147
00:06:47,660 --> 00:06:49,650
there are some creative ways
to find money

148
00:06:49,650 --> 00:06:52,363
like asking sponsors to cover expenses.

149
00:06:53,930 --> 00:06:57,430
I work with media producers,
corporations,

150
00:06:57,430 --> 00:07:01,780
educational institutions,
business owners, event organizers,

151
00:07:01,780 --> 00:07:05,930
and I help make their audio,
video, and events accessible

152
00:07:05,930 --> 00:07:08,200
at a reasonable cost.

153
00:07:08,200 --> 00:07:11,440
There's some guidance and information
along with my personal story

154
00:07:11,440 --> 00:07:13,660
that can be found in my book.

155
00:07:13,660 --> 00:07:16,480
If nobody asks for speech to text access,

156
00:07:16,480 --> 00:07:18,580
that does not mean that there's no demand.

157
00:07:19,420 --> 00:07:23,220
It needs to be provided
for all types of aural information

158
00:07:23,220 --> 00:07:25,800
even without a prior request.

159
00:07:25,800 --> 00:07:28,800
High quality, same language
captions and transcripts

160
00:07:28,800 --> 00:07:29,900
are good for business

161
00:07:29,900 --> 00:07:32,700
and a necessity for millions of people.

162
00:07:32,700 --> 00:07:34,914
Let's think outside the ears!

163
00:07:34,914 --> 00:07:35,747
(Laughter)

164
00:07:35,747 --> 00:07:37,610
Let's caption everything!

165
00:07:37,610 --> 00:07:39,840
Get your message to more people.

166
00:07:39,840 --> 00:07:40,822
Thank you.

167
00:07:40,822 --> 00:07:42,456
(Applause)