Rupal Patel: Synthetic voices, as unique as fingerprints

111,845 views ใƒป 2014-02-13

TED


์•„๋ž˜ ์˜๋ฌธ์ž๋ง‰์„ ๋”๋ธ”ํด๋ฆญํ•˜์‹œ๋ฉด ์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค.

๋ฒˆ์—ญ: K Bang ๊ฒ€ํ† : Jenny Lee
00:12
I'd like to talk today
0
12719
1490
์˜ค๋Š˜ ์ œ๊ฐ€ ๋ง์”€๋“œ๋ฆฌ๊ณ ์ž ํ•˜๋Š” ๊ฒƒ์€
00:14
about a powerful and fundamental aspect
1
14209
2927
์šฐ๋ฆฌ ์ž์‹ ์˜ ๊ฐ•๋ ฅํ•˜๊ณ ๋„ ๊ธฐ๋ณธ์ ์ธ
00:17
of who we are: our voice.
2
17136
3598
ํ•œ ์ธก๋ฉด์— ๋Œ€ํ•œ ๊ฒƒ์ž…๋‹ˆ๋‹ค. ๋ฐ”๋กœ ์šฐ๋ฆฌ์˜ ๋ชฉ์†Œ๋ฆฌ์ฃ .
00:20
Each one of us has a unique voiceprint
3
20734
2746
์šฐ๋ฆฌ ๋ชจ๋‘๋Š” ๋…ํŠนํ•œ ์„ฑ๋ฌธ์„ ๊ฐ€์ง€๊ณ  ์žˆ๋Š”๋ฐ์š”.
00:23
that reflects our age, our size,
4
23480
2289
์ด ์„ฑ๋ฌธ์€ ์šฐ๋ฆฌ์˜ ๋‚˜์ด, ํฌ๊ธฐ,
00:25
even our lifestyle and personality.
5
25769
3237
์‹ฌ์ง€์–ด ์ƒํ™œ ์–‘์‹์ด๋‚˜ ์„ฑ๊ฒฉ๊นŒ์ง€๋„ ๋“œ๋Ÿฌ๋‚ด์ค๋‹ˆ๋‹ค.
00:29
In the words of the poet Longfellow,
6
29006
2142
์‹œ์ธ ๋กฑํŽ ๋กœ์šฐ๋Š”
00:31
"the human voice is the organ of the soul."
7
31148
3870
"์ธ๊ฐ„์˜ ๋ชฉ์†Œ๋ฆฌ๋Š” ์˜ํ˜ผ์˜ ์˜ค๋ฅด๊ฐ„์ด๋‹ค." ๋ผ๊ณ  ๊นŒ์ง€ ํ–ˆ์Šต๋‹ˆ๋‹ค.
00:35
As a speech scientist, I'm fascinated
8
35018
2747
์Œ์„ฑํ•™์ž๋กœ์„œ ์ €๋Š” ์Œ์„ฑ์ด
00:37
by how the voice is produced,
9
37765
1829
์–ด๋–ป๊ฒŒ ๋งŒ๋“ค์–ด์ง€๋Š”์ง€์— ๋งค๋ฃŒ๋˜์–ด ์žˆ๊ณ 
00:39
and I have an idea for how it can be engineered.
10
39594
3658
๊ณตํ•™์ ์œผ๋กœ ์–ด๋–ป๊ฒŒ ๋งŒ๋“ค์–ด์งˆ ์ˆ˜ ์žˆ๋Š”์ง€์— ๋Œ€ํ•ด ์•Œ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
00:43
That's what I'd like to share with you.
11
43252
2210
์ด๊ฒŒ ์˜ค๋Š˜ ์ œ๊ฐ€ ์—ฌ๋Ÿฌ๋ถ„๊ณผ ๋“œ๋ฆด ์ด์•ผ๊ธฐ์˜ ์ฃผ์ œ์ž…๋‹ˆ๋‹ค.
00:45
I'm going to start by playing you a sample
12
45462
1814
์—ฌ๋Ÿฌ๋ถ„์ด ์•Œ์•„๋“ค์œผ์‹ค๋งŒํ•œ ๋ชฉ์†Œ๋ฆฌ๋กœ ์˜ˆ์‹œ๋ฅผ ๋“ค๋ฉฐ
00:47
of a voice that you may recognize.
13
47276
1871
์‹œ์ž‘ํ•ด ๋ณด๊ฒ ์Šต๋‹ˆ๋‹ค.
00:49
(Recording) Stephen Hawking: "I would have thought
14
49147
1304
(๋…น์Œ) ์Šคํ‹ฐ๋ธ ํ˜ธํ‚น: "์ œ๊ฐ€ ์ƒ๊ฐํ•˜๋Š” ๋ฐ”๋กœ๋Š”
00:50
it was fairly obvious what I meant."
15
50451
2749
์ œ ์˜๋„๊ฐ€ ์•„์ฃผ ๋ช…ํ™•ํ•ฉ๋‹ˆ๋‹ค."
00:53
Rupal Patel: That was the voice
16
53200
1280
๋ŸฌํŒฐ ํŒจํ…”: ์ €๊ฑด ๋ฐ”๋กœ
00:54
of Professor Stephen Hawking.
17
54480
2086
์Šคํ‹ฐ๋ธ ํ˜ธํ‚น ๊ต์ˆ˜์˜ ๋ชฉ์†Œ๋ฆฌ์˜€์Šต๋‹ˆ๋‹ค.
00:56
What you may not know is that same voice
18
56566
3849
์—ฌ๋Ÿฌ๋ถ„์ด ํฅ๋ฏธ๋กญ๊ฒŒ ๋Š๋ผ์‹ค๋งŒํ•œ ์ ์€,
01:00
may also be used by this little girl
19
60415
2478
์ด ์–ด๋ฆฐ ์†Œ๋…€๊ฐ€ ๋˜‘๊ฐ™์€ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์‚ฌ์šฉํ•  ์ˆ˜๋„ ์žˆ๋‹ค๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
01:02
who is unable to speak
20
62893
1697
์ด ์•„์ด๋Š” ์‹ ๊ฒฝํ•™์  ๋ฌธ์ œ๋กœ ์ธํ•ด
01:04
because of a neurological condition.
21
64590
2597
๋ง์„ ํ•  ์ˆ˜๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค.
01:07
In fact, all of these individuals
22
67187
2068
์‚ฌ์‹ค ์ด๋Ÿฐ ๋ชจ๋“  ์‚ฌ๋žŒ๋“ค์€
01:09
may be using the same voice,
23
69255
2012
๋˜‘๊ฐ™์€ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์“ฐ๊ณ  ์žˆ์„์ง€๋„ ๋ชจ๋ฆ…๋‹ˆ๋‹ค.
01:11
and that's because there's only a few options available.
24
71267
3557
๊ทธ ์ด์œ ๋Š” ์„ ํƒ ๊ฐ€๋Šฅํ•œ ๋ชฉ์†Œ๋ฆฌ ์ข…๋ฅ˜๊ฐ€ ๊ทธ๋ฆฌ ๋งŽ์ง€ ์•Š๊ธฐ ๋•Œ๋ฌธ์ด์—์š”.
01:14
In the U.S. alone, there are 2.5 million Americans
25
74824
4317
๋ฏธ๊ตญ์—๋งŒ ๋งํ•˜์ง€ ๋ชปํ•˜๋Š” ์‚ฌ๋žŒ์ด 250๋งŒ์ด๋‚˜
01:19
who are unable to speak,
26
79141
1610
๋ฉ๋‹ˆ๋‹ค.
01:20
and many of whom use computerized devices
27
80751
2622
๊ทธ๋“ค ์ค‘ ๋งŽ์€ ์‚ฌ๋žŒ๋“ค์€ ์ปดํ“จํ„ฐํ™”๋œ ๊ธฐ๊ธฐ๋ฅผ ์ด์šฉํ•˜์—ฌ
01:23
to communicate.
28
83373
1522
์˜์‚ฌ ์†Œํ†ต์„ ํ•ฉ๋‹ˆ๋‹ค.
01:24
Now that's millions of people worldwide
29
84895
3479
๊ทธ๋ ‡๋‹ค๋ฉด ์„ธ๊ณ„์ ์œผ๋กœ๋Š” ์ˆ˜๋ฐฑ๋งŒ์˜ ์‚ฌ๋žŒ๋“ค์ด
01:28
who are using generic voices,
30
88374
1652
๊ณตํ†ต์ ์ธ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์‚ฌ์šฉํ•˜๋Š”๊ฑฐ๊ฒ ์ฃ .
01:30
including Professor Hawking,
31
90026
1446
ํ˜ธํ‚น ๊ต์ˆ˜๋ฅผ ํฌํ•จํ•ด์„œ ๋ง์ž…๋‹ˆ๋‹ค.
01:31
who uses an American-accented voice.
32
91472
4833
๊ทธ ๋ถ„์€ ๋ฏธ๊ตญ ์–ต์–‘์„ ๊ฐ€์ง„ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์‚ฌ์šฉํ•ฉ๋‹ˆ๋‹ค.
01:36
This lack of individuation of the synthetic voice
33
96305
3328
์ด๋ ‡๋“ฏ ๊ธฐ๊ณ„ ์Œ์„ฑ์˜ ๊ฐœ๋ณ„ํ™”๊ฐ€ ๋ถ€์กฑํ•˜๋‹ค๋Š” ๊ฒƒ,
01:39
really hit home
34
99633
1416
์ด๊ฒŒ ๋ฐ”๋กœ ๋ฌธ์ œ์˜ ํ•ต์‹ฌ์ž…๋‹ˆ๋‹ค.
01:41
when I was at an assistive technology conference
35
101049
2472
์ œ๊ฐ€ ๋ณด์กฐ ๊ธฐ๊ตฌ ํšŒ์˜์— ์ฐธ์„ํ–ˆ๋˜ ์ ์ด ์žˆ์Šต๋‹ˆ๋‹ค.
01:43
a few years ago,
36
103521
1850
๋ช‡ ํ•ด ์ „์ด์—ˆ์ฃ .
01:45
and I recall walking into an exhibit hall
37
105371
3604
๊ทธ๋•Œ ์ €๋Š” ์ „์‹œ๊ด€์„ ๊ฑธ์–ด๊ฐ€๋‹ค๊ฐ€
01:48
and seeing a little girl and a grown man
38
108975
3044
์–ด๋ฆฐ ์†Œ๋…€์™€ ์„ฑ์ธ์ด ๊ฐ์ž์˜ ๊ธฐ๊ตฌ๋ฅผ ์‚ฌ์šฉํ•˜์—ฌ
01:52
having a conversation using their devices,
39
112019
2916
๋Œ€ํ™”ํ•˜๋Š” ๊ฒƒ์„ ๋ณด์•˜์Šต๋‹ˆ๋‹ค.
01:54
different devices, but the same voice.
40
114935
4284
์„œ๋กœ ๋‹ค๋ฅธ ๊ธฐ๊ตฌ์˜€์ง€๋งŒ ๋ชฉ์†Œ๋ฆฌ๋Š” ๋˜‘๊ฐ™์•˜์ฃ .
01:59
And I looked around and I saw this happening
41
119219
1909
๋‘˜๋Ÿฌ๋ณด๋‹ˆ ์ด๋Ÿฐ ์ผ์ด ์ œ ์ฃผ์œ„์—์„œ๋„ ๋นˆ๋ฒˆํ•˜๊ฒŒ ์ผ์–ด๋‚˜๊ณ  ์žˆ์—ˆ์ฃ .
02:01
all around me, literally hundreds of individuals
42
121128
4190
๋ง ๊ทธ๋Œ€๋กœ ์ˆ˜๋ฐฑ๋ช…์˜ ์‚ฌ๋žŒ๋“ค์ด
02:05
using a handful of voices,
43
125318
2738
๋‹จ ๋ช‡ ๊ฐ€์ง€ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ฐ–๊ณ  ์žˆ์—ˆ์Šต๋‹ˆ๋‹ค.
02:08
voices that didn't fit their bodies
44
128056
3091
๊ทธ๋“ค์˜ ์‹ ์ฒด๋‚˜ ์„ฑ๊ฒฉ๊ณผ๋Š”
02:11
or their personalities.
45
131147
2082
๋งž์ง€ ์•Š๋Š” ๋ชฉ์†Œ๋ฆฌ์˜€์–ด์š”.
02:13
We wouldn't dream of fitting a little girl
46
133229
2727
์šฐ๋ฆฌ๋Š” ์–ด๋ฆฐ ์†Œ๋…€์—๊ฒŒ
02:15
with the prosthetic limb of a grown man.
47
135956
3396
์„ฑ์ธ์˜ ๋ณด์กฐ ๊ธฐ๊ตฌ๋ฅผ ์‚ฌ์šฉ์‹œํ‚ฌ ์ƒ์ƒ๋„ ํ•˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค.
02:19
So why then the same prosthetic voice?
48
139352
3304
๊ทธ๋Ÿฐ๋ฐ ์™œ ๋ชฉ์†Œ๋ฆฌ๋Š” ๋˜‘๊ฐ™์€ ๊ฒƒ์„ ์‚ฌ์šฉํ• ๊นŒ์š”?
02:22
It really struck me,
49
142656
1291
์ด๊ฑด ์ •๋ง ์ œ๊ฒŒ ํฐ ์ถฉ๊ฒฉ์ด์—ˆ๊ณ 
02:23
and I wanted to do something about this.
50
143947
3151
์ €๋Š” ๋ญ”๊ฐ€ ํ•ด์•ผ๊ฒ ๋‹ค๋Š” ์ƒ๊ฐ์ด ๋“ค์—ˆ์Šต๋‹ˆ๋‹ค.
02:27
I'm going to play you now a sample
51
147098
1953
์—ฌ๋Ÿฌ๋ถ„๋“ค๊ป˜ ์–ด๋–ค ์‚ฌ๋žŒ์˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ
02:29
of someone who has, two people actually,
52
149051
3288
๋“ค๋ ค ๋“œ๋ฆดํ…๋ฐ์š”. ๋‘ ์‚ฌ๋žŒ์˜ ๋ชฉ์†Œ๋ฆฌ์ž…๋‹ˆ๋‹ค.
02:32
who have severe speech disorders.
53
152339
1768
์ด ๋‘ ์‚ฌ๋žŒ์—๊ฒŒ๋Š” ์‹ฌ๊ฐํ•œ ์œก์„ฑ ์žฅ์• ๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค.
02:34
I want you to take a listen to how they sound.
54
154107
3230
์–ด๋–ป๊ฒŒ ๋“ค๋ฆฌ๋Š”์ง€ ํ•œ๋ฒˆ ๋“ค์–ด๋ณด์„ธ์š”.
02:37
They're saying the same utterance.
55
157337
2357
๋‘ ์‚ฌ๋žŒ์€ ๋˜‘๊ฐ™์€ ๋ง์„ ํ•˜๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
02:39
(First voice)
56
159694
2432
(์ฒซ๋ฒˆ์งธ ์Œ์„ฑ)
02:42
(Second voice)
57
162126
3617
(๋‘๋ฒˆ์งธ ์Œ์„ฑ)
02:45
You probably didn't understand what they said,
58
165743
2412
์—ฌ๋Ÿฌ๋ถ„๋“ค์€ ์•„๋งˆ ๋‘ ์‚ฌ๋žŒ์ด ์–ด๋–ค ๋ง์„ ๋‚˜๋ˆด๋Š”์ง€ ๋ชจ๋ฅด์‹ค ๊ฒ๋‹ˆ๋‹ค.
02:48
but I hope that you heard
59
168155
1854
ํ•˜์ง€๋งŒ ์ €๋Š” ์—ฌ๋Ÿฌ๋ถ„๋“ค์ด ๊ฐ๊ฐ์˜
02:50
their unique vocal identities.
60
170009
4283
๋…ํŠนํ•œ ์Œ์„ฑ์„ ๋“ค์œผ์…จ๊ธฐ๋ฅผ ๋ฐ”๋ž๋‹ˆ๋‹ค.
02:54
So what I wanted to do next is,
61
174292
2813
๊ทธ๋ž˜์„œ ์ œ๊ฐ€ ๊ทธ ๋‹ค์Œ์œผ๋กœ ํ•˜๊ณ ์ž ํ•ด๋˜ ๊ฒŒ ์žˆ์Šต๋‹ˆ๋‹ค.
02:57
I wanted to find out how we could harness
62
177105
2384
์ €๋Š” ์ด๋ ‡๊ฒŒ ๋‚ด์žฌ๋œ ๋ฐœ์„ฑ ๋Šฅ๋ ฅ์„ ์–ด๋–ป๊ฒŒ ์ด์šฉํ•  ์ˆ˜ ์žˆ๋Š”์ง€
02:59
these residual vocal abilities
63
179489
1821
์•Œ์•„๋‚ด๊ณ  ์‹ถ์—ˆ์Šต๋‹ˆ๋‹ค.
03:01
and build a technology
64
181310
2016
๊ทธ๋ฆฌ๊ณ ๋Š” ๊ธฐ์ˆ ์„ ์ด์šฉํ•ด
03:03
that could be customized for them,
65
183326
2143
๊ฐ์ž์˜ ๋ชฉ์†Œ๋ฆฌ๋กœ ๊ฐœ๋ณ„ํ™”ํ•  ์ˆ˜ ์žˆ๊ธฐ๋ฅผ ํฌ๋งํ–ˆ์Šต๋‹ˆ๋‹ค.
03:05
voices that could be customized for them.
66
185469
2429
๊ฐ์ž์˜ ํŠน์„ฑ์„ ์ง€๋‹Œ ๋ชฉ์†Œ๋ฆฌ๋กœ ๋ง์ด์—์š”.
03:07
So I reached out to my collaborator, Tim Bunnell.
67
187898
2685
์ €๋Š” ์ œ ๋™๋ฃŒ์ธ ํŒ€ ๋ฒ„๋„ฌ์„ ์ฐพ์•„๊ฐ”์Šต๋‹ˆ๋‹ค.
03:10
Dr. Bunnell is an expert in speech synthesis,
68
190583
3063
๋ฒ„๋„ฌ ๋ฐ•์‚ฌ๋‹˜์€ ์Œ์„ฑ ๋ถ„์„ ์ „๋ฌธ๊ฐ€์ธ๋ฐ
03:13
and what he'd been doing is building
69
193646
2033
๊ทธ๋ถ„์ด ํ•˜๊ณ  ๊ณ„์‹œ๋˜ ์ผ์€
03:15
personalized voices for people
70
195679
1881
์‚ฌ๋žŒ๋“ค์˜ ๊ฐœ๋ณ„์ ์ธ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋งŒ๋“œ๋Š” ์ผ์ด์—ˆ์–ด์š”.
03:17
by putting together
71
197560
2097
๋ฏธ๋ฆฌ ๋…น์Œ๋œ
03:19
pre-recorded samples of their voice
72
199657
2150
์‚ฌ๋žŒ๋“ค์˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์กฐํ•ฉํ•˜์—ฌ
03:21
and reconstructing a voice for them.
73
201807
2879
๊ฐœ๊ฐœ์ธ์„ ์œ„ํ•œ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋‹ค์‹œ ๋งŒ๋“ค์–ด ๋‚ด๋Š” ์ผ์ด์—ˆ์ฃ .
03:24
These are people who had lost their voice
74
204686
1712
์ด ๋ถ„๋“ค์€ ํ›„์ฒœ์ ์œผ๋กœ ์ž์‹ ์˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ
03:26
later in life.
75
206398
1911
์žƒ์€ ์‚ฌ๋žŒ๋“ค์ด์—์š”.
03:28
We didn't have the luxury
76
208309
1394
์šฐ๋ฆฌ์—๊ฒ ์„ ์ฒœ์ ์œผ๋กœ
03:29
of pre-recorded samples of speech
77
209703
1774
์Œ์„ฑ ์žฅ์• ๋ฅผ ๊ฐ€์ง€๊ณ  ํƒœ์–ด๋‚œ ์‚ฌ๋žŒ๋“ค์ด
03:31
for those born with speech disorder.
78
211477
2292
์ด๋ฏธ ๋…น์Œํ•œ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ฐ€์ง€๋Š” ์‚ฌ์น˜ ๊ฐ™์€ ๊ฒƒ์€ ์—†์—ˆ์Šต๋‹ˆ๋‹ค.
03:33
But I thought, there had to be a way
79
213769
2537
ํ•˜์ง€๋งŒ ์•„๋ฌด๋ฆฌ ์ ์€ ๊ฒƒ์ด๋ผ๋„ ๋‚จ์•„์žˆ์œผ๋ฉด ๊ทธ๊ฒƒ์„ ํ†ตํ•ด
03:36
to reverse engineer a voice
80
216306
1944
๊ณตํ•™์ ์œผ๋กœ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋˜์‚ด๋ ค ๋‚ผ ์ˆ˜ ์žˆ๋Š”
03:38
from whatever little is left over.
81
218250
2291
๋ฐฉ๋ฒ•์ด ์žˆ์„๊ฑฐ๋ผ๊ณ  ์ƒ๊ฐํ–ˆ์–ด์š”.
03:40
So we decided to do exactly that.
82
220541
2714
๋ฐ”๋กœ ๊ทธ ๋ฐฉ๋ฒ•์„ ์จ๋ณด๊ธฐ๋กœ ํ•œ๊ฒ๋‹ˆ๋‹ค.
03:43
We set out with a little bit of funding from the National Science Foundation,
83
223255
3403
์šฐ๋ฆฌ๋Š” ๊ตญ๋ฆฝ๊ณผํ•™์žฌ๋‹จ(NSF)๋กœ ๋ถ€ํ„ฐ ๋ฐ›์€ ์•ฝ๊ฐ„์˜ ์žฌ์ • ์ง€์›์„ ๊ฐ€์ง€๊ณ 
03:46
to create custom-crafted voices that captured
84
226658
3565
๊ฐ์ž์˜ ์œ ์ผํ•œ ์Œ์„ฑ์  ํŠน์„ฑ์„ ๊ฐ–๋„๋ก
03:50
their unique vocal identities.
85
230223
1536
๋งž์ถค ์ œ์ž‘ํ˜• ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋งŒ๋“ค๊ธฐ ์‹œ์ž‘ํ–ˆ์Šต๋‹ˆ๋‹ค.
03:51
We call this project VocaliD, or vocal I.D.,
86
231759
3203
์ €ํฌ๋Š” ์ด ํ”„๋กœ์ ํŠธ๋ฅผ ๋ณด์ปฌ์•„์ด๋””( vocal I.D.)๋ผ๊ณ  ๋ถˆ๋ €์Šต๋‹ˆ๋‹ค.
03:54
for vocal identity.
87
234962
2033
vocal identitiy์—์„œ ๋”ฐ์™”์ฃ .
03:56
Now before I get into the details of how
88
236995
2674
์ €ํฌ๊ฐ€ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์–ด๋–ป๊ฒŒ ๋งŒ๋“ค์—ˆ๋Š”์ง€์— ๋Œ€ํ•œ
03:59
the voice is made and let you listen to it,
89
239669
2048
์ž์„ธํ•œ ์ด์•ผ๊ธฐ๋‚˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์—ฌ๋Ÿฌ๋ถ„๊ป˜ ๋“ค๋ ค๋“œ๋ฆฌ๊ธฐ ์ „์—
04:01
I need to give you a real quick speech science lesson. Okay?
90
241717
3350
์—ฌ๋Ÿฌ๋ถ„๊ป˜ ์•„์ฃผ ์งง์€ ์Œ์„ฑ ๊ณผํ•™ ๊ต์œก์„ ํ•ด์•ผํ•  ํ•„์š”๊ฐ€ ์žˆ์Šต๋‹ˆ๋‹ค, ๊ดœ์ฐฎ๊ฒ ์ฃ ?
04:05
So first, we know that the voice is changing
91
245067
3159
์šฐ์„ , ์šฐ๋ฆฌ๋Š” ์‚ฌ๋žŒ์˜ ๋ชฉ์†Œ๋ฆฌ๊ฐ€ ๊ทธ ์ฃผ์ธ์ด ์„ฑ์žฅํ•จ์— ๋”ฐ๋ผ
04:08
dramatically over the course of development.
92
248226
2854
๊ธ‰๊ฒฉํ•˜๊ฒŒ ๋ณ€ํ•œ๋‹ค๋Š” ์‚ฌ์‹ค์„ ์•Œ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.
04:11
Children sound different from teens
93
251080
2090
์–ด๋ฆฐ ์•„์ด์ผ ๋•Œ์˜ ์Œ์„ฑ์€ ์ฒญ์†Œ๋…„๊ธฐ์˜ ์Œ์„ฑ๊ณผ ๋‹ค๋ฅด๊ณ 
04:13
who sound different from adults.
94
253170
1463
๋˜ ๊ทธ๊ฑด ์„ฑ์ธ์ด ๋˜์—ˆ์„ ๋•Œ์™€๋„ ๋‹ค๋ฆ…๋‹ˆ๋‹ค.
04:14
We've all experienced this.
95
254633
2642
์šฐ๋ฆฌ ๋ชจ๋‘ ์ด๋Ÿฐ ๊ฒฝํ—˜์„ ํ•ด๋ณธ ์ ์ด ์žˆ์ง€์š”.
04:17
Fact number two is that speech
96
257275
3363
๋‘๋ฒˆ์งธ ์‚ฌ์‹ค์€ ์Œ์„ฑ์ด๋ž€ ๊ฒƒ์ด
04:20
is a combination of the source,
97
260638
2553
๊ฐ์ž์˜ ํ›„๋‘์—์„œ ๋ฐœ์ƒํ•˜๋Š”
04:23
which is the vibrations generated by your voice box,
98
263191
3479
์ง„๋™, ์ฆ‰ ์Œ์›๊ณผ
04:26
which are then pushed through
99
266670
1939
๊ทธ ์Œ์›์ด ๋‚˜๋จธ์ง€ ์„ฑ๋„ ๋ถ€๋ถ„์„ ํ†ตํ•ด ๋ฐ€๋ ค๋‚˜๊ฐ€๋ฉฐ ์ผ์–ด๋‚˜๋Š” ํ˜„์ƒ์˜
04:28
the rest of the vocal tract.
100
268609
2437
์กฐํ™”๋กœ ๋งŒ๋“ค์–ด์ง„๋‹ค๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.
04:31
These are the chambers of your head and neck
101
271046
2484
์ด๋Ÿฐ ๊ฒƒ๋“ค์€ ์‚ฌ๋žŒ์˜ ๋จธ๋ฆฌ์™€ ๋ชฉ์— ์žˆ๋Š” ๊ณต๊ฐ„์ธ๋ฐ์š”.
04:33
that vibrate,
102
273530
1239
์ด๋“ค์ด ์ง„๋™ํ•˜๊ณ 
04:34
and they actually filter that source sound
103
274769
2110
์Œ์„ฑ์›์„ ํˆฌ๊ณผ์‹œ์ผœ
04:36
to produce consonants and vowels.
104
276879
2537
์ž์Œ๊ณผ ๋ชจ์Œ์„ ๋งŒ๋“ค์–ด ๋ƒ…๋‹ˆ๋‹ค.
04:39
So the combination of source and filter
105
279416
3860
์ฆ‰, ์Œ์„ฑ์›๊ณผ ํˆฌ๊ณผ ๊ณผ์ •์˜ ์กฐํ•ฉ์œผ๋กœ
04:43
is how we produce speech.
106
283276
2630
์šฐ๋ฆฌ๊ฐ€ ๋งํ•˜๊ฒŒ ๋˜๋Š” ๊ฒ๋‹ˆ๋‹ค.
04:45
And that happens in one individual.
107
285906
3026
๊ทธ๊ฒƒ์ด ๋ชจ๋“  ๊ฐœ๊ฐœ์ธ์—๊ฒŒ์„œ ์ผ์–ด๋‚˜์ฃ .
04:48
Now I told you earlier that I'd spent
108
288932
2626
์•ž์—์„œ ๋ง์”€๋“œ๋ฆฐ๋Œ€๋กœ
04:51
a good part of my career
109
291558
2025
์ €๋Š” ์ œ ์ผ์ƒ์˜ ์ƒ๋‹น ๋ถ€๋ถ„์„
04:53
understanding and studying
110
293583
2453
์‹ฌ๊ฐํ•œ ์Œ์„ฑ ์žฅ์• ๋ฅผ ๊ฐ€์ง„ ์‚ฌ๋žŒ๋“ค์˜
04:56
the source characteristics of people
111
296036
1958
์Œ์„ฑ์›์ด ๊ฐ€์ง„ ํŠน์„ฑ์„ ์ดํ•ดํ•˜๊ณ 
04:57
with severe speech disorder,
112
297994
2301
์—ฐ๊ตฌํ•ด ์™”์Šต๋‹ˆ๋‹ค.
05:00
and what I've found
113
300295
1465
์ œ๊ฐ€ ์•Œ์•„๋‚ธ ๊ฒƒ์€
05:01
is that even though their filters were impaired,
114
301760
3366
๊ทธ๋“ค์˜ ํˆฌ๊ณผ ์žฅ์น˜๊ฐ€ ์†์ƒ๋˜์–ด๋„
05:05
they were able to modulate their source:
115
305126
2961
์Œ์„ฑ์›์„ ์กฐ์ ˆํ•  ์ˆ˜ ์žˆ๋‹ค๋Š” ์‚ฌ์‹ค์ž…๋‹ˆ๋‹ค:
05:08
the pitch, the loudness, the tempo of their voice.
116
308087
3262
๋ชฉ์†Œ๋ฆฌ์˜ ๋†’๋‚ฎ์ด, ํฌ๊ธฐ, ๋น ๋ฅด๊ฒŒ ๊ฐ™์€ ๊ฒƒ๋“ค์ด์š”.
05:11
These are called prosody, and I've been documenting for years
117
311349
3368
์ด๋Ÿฐ ๊ฒƒ๋“ค์„ ์šด์œจ ์ฒด๊ณ„๋ผ๊ณ  ํ•˜๋Š”๋ฐ, ์ €๋Š” ์ด๋Ÿฐ ๋ถ„๋“ค์˜ ์šด์œจ์  ๋Šฅ๋ ฅ์ด
05:14
that the prosodic abilities of these individuals
118
314717
2277
๋ณด์กด๋œ๋‹ค๋Š” ์ ์„ ์—ฌ๋Ÿฌ ํ•ด์— ๊ฑธ์ณ
05:16
are preserved.
119
316994
1575
์กฐ์‚ฌํ•ด ์™”์Šต๋‹ˆ๋‹ค.
05:18
So when I realized that those same cues
120
318569
4087
๊ทธ๋ž˜์„œ ์ด๋Ÿฐ ๊ฒƒ๋“ค์ด
05:22
are also important for speaker identity,
121
322656
2769
๋งํ•˜๋Š” ์‚ฌ๋žŒ์˜ ํŠน์„ฑ์— ์ค‘์š”ํ•˜๋‹ค๋Š” ๊ฒƒ์— ์ƒ๊ฐ์ด ๋ฏธ์น˜์ž
05:25
I had this idea.
122
325425
2015
๋ฐ”๋กœ ์ด๋Ÿฐ ์•„์ด๋””์–ด๊ฐ€ ๋– ์˜ฌ๋ž์Šต๋‹ˆ๋‹ค.
05:27
Why don't we take the source
123
327440
2516
์›ํ•˜๋Š” ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ฐ€์ง„ ์‚ฌ๋žŒ๋“ค์˜
05:29
from the person we want the voice to sound like,
124
329956
2213
์Œ์„ฑ์›์„ ๊ฐ€์ ธ์˜ค๋ฉด ์–ด๋–จ๊นŒํ•˜๋Š” ๊ฒƒ์ด์—ˆ์ฃ .
05:32
because it's preserved,
125
332169
1463
์Œ์„ฑ์›์€ ๋ณด์กด๋˜๋‹ˆ๊นŒ์š”.
05:33
and borrow the filter
126
333632
2135
๊ทธ๋ฆฌ๊ณ  ์ด ํˆฌ๊ณผ ๊ธฐ๊ด€์ด ์Œ์„ฑ์„ ๊ตฌ๋ถ„์ง€์„ ์ˆ˜ ์žˆ์œผ๋‹ˆ๊นŒ
05:35
from someone about the same age and size,
127
335767
3229
๋น„์Šทํ•œ ๋‚˜์ด์™€ ์‹ ์ฒด ํฌ๊ธฐ๋ฅผ ๊ฐ€์ง„ ์‚ฌ๋žŒ์˜
05:39
because they can articulate speech,
128
339011
2407
ํˆฌ๊ณผ ๊ธฐ๊ฐ„์„ ์ฐจ์šฉํ•˜์—ฌ
05:41
and then mix them?
129
341418
1791
์Œ์„ฑ์›๊ณผ ํ˜ผํ•ฉํ•˜๋ฉด ์–ด๋–จ๊นŒํ•˜๋Š” ์ƒ๊ฐ์ด์—ˆ์Šต๋‹ˆ๋‹ค.
05:43
Because when we mix them,
130
343209
1787
์™œ๋ƒํ•˜๋ฉด ๊ทธ๊ฒƒ๋“ค์„ ํ•ฉ์„ฑํ•˜๋ฉด
05:44
we can get a voice that's as clear
131
344996
1698
๋Œ€๋ฆฌ์ž์˜ ๋ชฉ์†Œ๋ฆฌ ๋งŒํผ์ด๋‚˜
05:46
as our surrogate talker --
132
346694
1754
๋ช…ํ™•ํ•œ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์–ป์„ ์ˆ˜ ์žˆ์œผ๋‹ˆ๊นŒ์š”.
05:48
that's the person we borrowed the filter fromโ€”
133
348448
2595
ํˆฌ๊ณผ ๊ธฐ๊ด€์„ ์ฐจ์šฉํ•ด ์ค€ ์‚ฌ๋žŒ์ด ๋Œ€๋ฆฌ์ž๊ฐ€ ๋˜๋Š” ๊ฒ๋‹ˆ๋‹ค.
05:51
and is similar in identity to our target talker.
134
351043
4649
๊ทธ๋ฆฌ๊ณ  ๊ทธ ๋ชฉ์†Œ๋ฆฌ์˜ ํŠน์„ฑ์€ ์œก์„ฑ์„ ๋‚ด๊ณ ์ž ํ•˜๋Š” ์‚ฌ๋žŒ์˜ ๋ชฉ์†Œ๋ฆฌ์™€ ๋น„์Šทํ•˜๊ฒŒ ๋˜๋Š”๊ฑฐ์ฃ .
05:55
It's that simple.
135
355692
1427
์•„์ฃผ ๊ฐ„๋‹จํ•˜์ฃ .
05:57
That's the science behind what we're doing.
136
357119
2934
์ด๊ฒŒ ๋ฐ”๋กœ ์ €ํฌ๊ฐ€ ํ•˜๊ณ  ์žˆ๋Š” ์ผ์˜ ๊ณผํ•™์  ์›๋ฆฌ๋ผ๊ณ  ํ•  ์ˆ˜ ์žˆ์Šต๋‹ˆ๋‹ค.
06:00
So once you have that in mind,
137
360053
3533
๊ทธ๋Ÿผ ์ด๋Ÿฐ ๊ฒƒ์„ ์—ผ๋‘์— ๋‘์—ˆ์„ ๋•Œ,
06:03
how do you go about building this voice?
138
363586
2258
์ด๋Ÿฐ ๋ชฉ์†Œ๋ฆฌ๋Š” ์–ด๋–ป๊ฒŒ ๋งŒ๋“œ๋Š” ๊ฑธ๊นŒ์š”?
06:05
Well, you have to find someone
139
365844
1480
์šฐ์„  ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋นŒ๋ ค์ค„ ์‚ฌ๋žŒ์„
06:07
who is willing to be a surrogate.
140
367324
2400
์ฐพ์•„์•ผ ํ•ฉ๋‹ˆ๋‹ค.
06:09
It's not such an ominous thing.
141
369724
2264
๊ทธ๊ฑด ๊ทธ๋ฆฌ ์–ด๋ ค์šด ์ผ์ด ์•„๋‹ˆ์ฃ .
06:11
Being a surrogate donor
142
371988
1523
๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋นŒ๋ ค์ฃผ๋Š” ๋ฐ์—๋Š”
06:13
only requires you to say a few hundred
143
373511
2788
๋ช‡ ๋ฐฑ์—์„œ ๋ช‡ ์ฒœ๊ฐœ์ฏค์˜
06:16
to a few thousand utterances.
144
376299
2242
์†Œ๋ฆฌ๋งŒ ๋‚ด์ฃผ๋ฉด ๋ฉ๋‹ˆ๋‹ค.
06:18
The process goes something like this.
145
378541
2003
๊ทธ ๊ณผ์ •์€ ์ด๋ ‡์Šต๋‹ˆ๋‹ค.
06:20
(Video) Voice: Things happen in pairs.
146
380544
2190
(์˜์ƒ) ๋ชฉ์†Œ๋ฆฌ: ํ˜„์ƒ์€ ์ง์œผ๋กœ ์ผ์–ด๋‚œ๋‹ค.
06:22
I love to sleep.
147
382734
1925
๋‚˜๋Š” ์ž ์ž๋Š” ๊ฑธ ์ข‹์•„ํ•ด.
06:24
The sky is blue without clouds.
148
384659
3882
ํ•˜๋Š˜์ด ๊ตฌ๋ฆ„ ํ•œ์ ์—†์ด ๋ง‘๊ตฐ.
06:28
RP: Now she's going to go on like this
149
388541
2002
ํŽ˜ํ…”: ์ด์ œ ๊ทธ๋…€๋Š” ์ด๋ ‡๊ฒŒ
06:30
for about three to four hours,
150
390543
1919
์„œ๋„ˆ ์‹œ๊ฐ„์„ ๊ณ„์†ํ•ฉ๋‹ˆ๋‹ค.
06:32
and the idea is not for her to say everything
151
392462
3005
์š”์ ์€ ๊ทธ๋…€๊ฐ€ ์ด๋ ‡๊ฒŒ ๋ชจ๋“  ๋‹จ์–ด๋ฅผ
06:35
that the target is going to want to say,
152
395467
2045
๋งํ•˜๋Š” ๊ฒƒ์ด ์•„๋‹ˆ๋ผ
06:37
but the idea is to cover all the different combinations
153
397512
3395
์–ธ์–ด์—์„œ ์ผ์–ด๋‚  ์ˆ˜ ์žˆ๋Š”
06:40
of the sounds that occur in the language.
154
400907
3271
๋ชจ๋“  ์Œ์„ฑ์˜ ๋‹ค๋ฅธ ์กฐํ•ฉ๋“ค์„ ๋งŒ๋“ค์–ด ๋‚ด๋Š” ๊ฒƒ์ด์—์š”.
06:44
The more speech you have,
155
404178
1638
๋ฐœ์Œ์ƒ˜ํ”Œ์„ ๋” ๋งŽ์ด ํ™•๋ณดํ•  ์ˆ˜๋ก
06:45
the better sounding voice you're going to have.
156
405816
2305
๋” ์ข‹์€ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ฐ–๊ฒŒ ๋ฉ๋‹ˆ๋‹ค.
06:48
Once you have those recordings,
157
408121
1673
์ด๊ฑธ ๋ชจ๋‘ ๋…น์Œํ•œ ๋‹ค์Œ์—
06:49
what we need to do
158
409794
1413
ํ•ด์•ผ ํ•  ์ผ์€
06:51
is we have to parse these recordings
159
411207
2718
์ด ๋…น์Œ์„ ๋ถ„์„ํ•˜์—ฌ
06:53
into little snippets of speech,
160
413925
2449
๋ง์— ์“ฐ์ด๋Š” ์ž‘์€ ์กฐ๊ฐ์œผ๋กœ
06:56
one- or two-sound combinations,
161
416374
2337
๊ทธ๋ฆฌ๊ณ  ์ด๊ฑธ ๋‹ค์‹œ ํ•œ๋‘๊ฐ€์ง€ ์†Œ๋ฆฌ์˜ ์กฐํ•ฉ์œผ๋กœ ๋งŒ๋“œ๋Š”๋ฐ,
06:58
sometimes even whole words
162
418711
1883
๋•Œ๋ก  ์ „์ฒด ๋‹จ์–ด๋กœ
07:00
that start populating a dataset or a database.
163
420594
4516
์ž๋ฃŒ ์ง‘ํ•ฉ ํ˜น์€ ๋ฐ์ดํ„ฐ๋ฒ ์ด์Šค๋ฅผ ๋งŽ์ด ๋งŒ๋“ค์–ด๋ƒ…๋‹ˆ๋‹ค.
07:05
We're going to call this database a voice bank.
164
425110
3717
์ด๋Ÿฐ ์ž๋ฃŒ ์กฐํ•ฉ์„ ์Œ์„ฑ ์€ํ–‰์ด๋ผ๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
07:08
Now the power of the voice bank
165
428827
2096
์Œ์„ฑ ์€ํ–‰์˜ ํž˜์€
07:10
is that from this voice bank,
166
430923
2014
๊ทธ๋กœ๋ถ€ํ„ฐ
07:12
we can now say any new utterance,
167
432937
2011
์–ด๋–ค ์ƒˆ๋กœ์šด ์Œ์„ฑ์ด๋ผ๋„ ๋งŒ๋“ค์–ด๋‚ผ ์ˆ˜ ์žˆ๋‹ค๋Š” ์ ์— ์žˆ์Šต๋‹ˆ๋‹ค.
07:14
like, "I love chocolate" --
168
434948
1424
--"๋‚˜๋Š” ์ตธ์ฝ”๋ ›์„ ์‚ฌ๋ž‘ํ•ด." ๊ฐ™์€ ๋ง๋„์š”.
07:16
everyone needs to be able to say thatโ€”
169
436372
1739
๋ˆ„๊ตฌ๋‚˜ ์ด ๋ง์€ ๊ผญ ํ•  ์ˆ˜ ์žˆ์–ด์•ผ ํ•˜์ž–์•„์š”. --
07:18
fish through that database
170
438111
1831
๊ทธ ์ž๋ฃŒ ์ง‘ํ•ฉ์—์„œ ๊ฑธ๋Ÿฌ๋‚ด
07:19
and find all the segments necessary
171
439942
1940
๊ทธ๋Ÿฐ ๋ฐœ์Œ์„ ํ•˜๋Š”๋ฐ ํ•„์š”ํ•œ
07:21
to say that utterance.
172
441882
1929
๋ชจ๋“  ๋‹จ์œ„๋ฅผ ์ฐพ์•„๋ƒ…๋‹ˆ๋‹ค.
07:23
(Video) Voice: I love chocolate.
173
443811
1789
(์˜์ƒ) ์Œ์„ฑ: ๋‚˜๋Š” ์ดˆ์ฝ”๋ ›์„ ์‚ฌ๋ž‘ํ•ด.
07:25
RP: So that's speech synthesis.
174
445600
1391
ํŒจํ…”: ์ด๊ฒƒ์ด ํ†ตํ•ฉ๋œ ์Œ์„ฑ์ž…๋‹ˆ๋‹ค.
07:26
It's called concatenative synthesis, and that's what we're using.
175
446991
2573
ํ”ํžˆ '๋ณตํ•ฉ์  ํ†ตํ•ฉ'์ด๋ผ๊ณ  ํ•˜๋Š”๋ฐ์š”. ์šฐ๋ฆฌ๊ฐ€ ์‚ฌ์šฉํ•˜๋Š” ์Œ์„ฑ์ด ๋ฐ”๋กœ ์ด๊ฑฐ์—์š”.
07:29
That's not the novel part.
176
449564
1533
์ด๊ฑด ํŠน๋ณ„ํžˆ ์ƒˆ๋กœ์šด ๊ฒƒ์€ ์•„๋‹™๋‹ˆ๋‹ค.
07:31
What's novel is how we make it sound
177
451097
2221
๋…์ฐฝ์ ์ธ ๋ถ€๋ถ„์€ ์šฐ๋ฆฌ๊ฐ€ ๊ทธ๊ฑธ ์ด์šฉํ•ด ์œก์„ฑ์„ ๋งŒ๋“œ๋Š” ๋ถ€๋ถ„์ด์—์š”.
07:33
like this young woman.
178
453318
1457
์ด ์ Š์€ ์—ฌ์„ฑ์ฒ˜๋Ÿผ์ด์š”.
07:34
This is Samantha.
179
454775
1524
์ด ์—ฌ์„ฑ๋ถ„์˜ ์ด๋ฆ„์€ ์‚ฌ๋งŒ๋‹ค๋ผ๊ณ  ํ•ด์š”.
07:36
I met her when she was nine,
180
456299
2346
์ €๋Š” ์ด ์•„์ด๊ฐ€ 9์‚ด๋•Œ ์ฒ˜์Œ ๋งŒ๋‚ฌ๋Š”๋ฐ
07:38
and since then, my team and I
181
458645
1897
๊ทธ ์ดํ›„๋กœ, ์ €ํฌ ํŒ€๊ณผ ์ €๋Š”
07:40
have been trying to build her a personalized voice.
182
460542
2714
์ด ์•„์ด๋งŒ์˜ ์Œ์„ฑ์„ ๋งŒ๋“ค๋ ค๊ณ  ๋…ธ๋ ฅํ–ˆ์Šต๋‹ˆ๋‹ค.
07:43
We first had to find a surrogate donor,
183
463256
3099
์ฒ˜์Œ์— ์šฐ๋ฆฌ๋Š” ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋นŒ๋ ค์ค„ ์‚ฌ๋žŒ์„ ์ฐพ์•„์•ผ๋งŒ ํ–ˆ์Šต๋‹ˆ๋‹ค.
07:46
and then we had to have Samantha
184
466355
1818
๊ทธ๋ฆฌ๊ณ ๋Š” ์‚ฌ๋งŒ๋‹ค๊ฐ€
07:48
produce some utterances.
185
468173
1929
๋ช‡๊ฐ€์ง€ ์†Œ๋ฆฌ๋ฅผ ๋‚ด๋„๋ก ํ–ˆ์ฃ .
07:50
What she can produce are mostly vowel-like sounds,
186
470102
2379
๊ทธ ์•„์ด๊ฐ€ ๋‚ผ ์ˆ˜ ์žˆ๋Š” ์†Œ๋ฆฌ๋Š” ๋Œ€๋ถ€๋ถ„ ๋ชจ์Œ์ฒ˜๋Ÿผ ๋“ค๋ฆฌ๋Š” ๊ฒƒ์ด์—ˆ์ง€๋งŒ
07:52
but that's enough for us to extract
187
472481
2479
์ €ํฌ๊ฐ€ ๊ทธ ์•„์ด์˜ ํŠน์ƒ‰์„ ์‚ด๋ฆฐ
07:54
her source characteristics.
188
474960
2285
์œก์„ฑ์„ ์ถ”์ถœํ•˜๊ธฐ์—๋Š” ์ถฉ๋ถ„ํ–ˆ์Šต๋‹ˆ๋‹ค.
07:57
What happens next is best described
189
477245
3271
๊ทธ ๋‹ค์Œ์œผ๋กœ ํ–ˆ๋˜ ์ผ์€
08:00
by my daughter's analogy. She's six.
190
480516
2767
์ œ ๋”ธ์•„์ด๊ฐ€ ํ–ˆ๋˜ ๊ฒƒ๊ณผ ๋˜‘๊ฐ™์•„์š”. ์ด์ œ 6์‚ด์ธ๋ฐ์š”.
08:03
She calls it mixing colors to paint voices.
191
483283
5422
๊ทธ ์• ๋Š” ์†Œ๋ฆฌ๋ฅผ ํ•ฉ์„ฑํ•˜๋Š” ๊ฒƒ์„ ๋ชฉ์†Œ๋ฆฌ์— ์ƒ‰์„ ์ž…ํžŒ๋‹ค๊ณ  ํ•ฉ๋‹ˆ๋‹ค.
08:08
It's beautiful. It's exactly that.
192
488705
2555
์•„์ฃผ ์•„๋ฆ„๋‹ค์šด ๋น„์œ ์ฃ . ๋”ฑ ๊ทธ๋ง์ด ๋งž์Šต๋‹ˆ๋‹ค.
08:11
Samantha's voice is like a concentrated sample
193
491260
2860
์‚ฌ๋งŒ๋‹ค์˜ ๋ชฉ์†Œ๋ฆฌ๋Š” ๋นจ๊ฐ•์ƒ‰ ์‹์šฉ์ƒ‰์†Œ๋ฅผ
08:14
of red food dye which we can infuse
194
494120
2609
๋†์ถ•ํ•ด ๋†“์€ ๊ฒƒ๊ณผ ๊ฐ™์•„์„œ
08:16
into the recordings of her surrogate
195
496729
2540
๊ทธ๊ฑธ ๊ธฐ์ฆ๋œ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋…น์Œํ•œ ๊ฒƒ์— ์ž…ํ˜€
08:19
to get a pink voice just like this.
196
499269
4387
์ด๋Ÿฐ ํ•‘ํฌ๋น› ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋งŒ๋“ค์–ด ๋ƒ…๋‹ˆ๋‹ค.
08:23
(Video) Samantha: Aaaaaah.
197
503656
4491
(์˜์ƒ) ์‚ฌ๋งˆ๋‹ค: ์•„~~
08:28
RP: So now, Samantha can say this.
198
508147
2808
ํŒจํ…”: ์ด์ œ ์‚ฌ๋งŒ๋‹ค๋Š” ์ด๋Ÿฐ ๋ง์„ ํ•  ์ˆ˜ ์žˆ์–ด์š”.
08:30
(Video) Samantha: This voice is only for me.
199
510955
3069
(์˜์ƒ) ์‚ฌ๋งŒ๋‹ค: ์ด ๋ชฉ์†Œ๋ฆฌ๋Š” ์˜ค์ง ๋‚˜๋ฅผ ์œ„ํ•œ ๋ชฉ์†Œ๋ฆฌ๋‹ค.
08:34
I can't wait to use my new voice with my friends.
200
514024
6305
์ƒˆ๋กœ ๋ฐ›์€ ๋ชฉ์†Œ๋ฆฌ๋กœ ์–ด์„œ ์นœ๊ตฌ๋“ค๊ณผ ์–˜๊ธฐํ•˜๊ณ  ์‹ถ์–ด์š”.
08:40
RP: Thank you. (Applause)
201
520329
6417
ํŒจํ…”: ๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค. (๋ฐ•์ˆ˜)
08:46
I'll never forget the gentle smile
202
526746
2333
์ €๋Š” ์‚ฌ๋งŒ๋‹ค๊ฐ€ ์ด ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์ฒ˜์Œ ๋“ค์—ˆ์„ ๋•Œ,
08:49
that spread across her face
203
529079
1902
์•„์ด์˜ ์–ผ๊ตด์— ํผ์ง€๋˜
08:50
when she heard that voice for the first time.
204
530981
3649
๋ถ€๋“œ๋Ÿฌ์šด ๋ฏธ์†Œ๋ฅผ ์ ˆ๋Œ€๋กœ ์žŠ์„ ์ˆ˜๊ฐ€ ์—†์Šต๋‹ˆ๋‹ค.
08:54
Now there's millions of people
205
534630
1882
์ „์„ธ๊ณ„์—๋Š” ์‚ฌ๋งŒ๋‹ค์™€ ๊ฐ™์€ ์‚ฌ๋žŒ๋“ค์ด
08:56
around the world like Samantha, millions,
206
536512
2833
์ˆ˜๋ฐฑ๋งŒ ๋ช…์ด๋‚˜ ์žˆ์–ด์š”. ์ˆ˜๋ฐฑ๋งŒ์ด์š”.
08:59
and we've only begun to scratch the surface.
207
539345
3440
์ €ํฌ๋Š” ์ด์ œ ๊ฒจ์šฐ ํ•œ๊ฑธ์Œ ๋‚ด๋”›๊ธฐ ์‹œ์ž‘ํ•œ ์ •๋„๊ตฌ์š”.
09:02
What we've done so far is we have
208
542785
1642
์ด์ œ๊นŒ์ง€ ์šฐ๋ฆฌ๊ฐ€ ํ•ด๋‚ธ ์ผ์€
09:04
a few surrogate talkers from around the U.S.
209
544427
3859
๋ฏธ๊ตญ์—์„œ ์ž์‹ ์˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ธฐ๋ถ€ํ•˜๊ฒ ๋‹ค๊ณ  ํ•ด์ค€
09:08
who have donated their voices,
210
548286
1507
๋ช‡๋ช‡์˜ ๋ด‰์‚ฌ์ž๋ฅผ ์ฐพ์•„๋ƒˆ๊ณ 
09:09
and we have been using those
211
549793
1928
์šฐ๋ฆฌ๋Š” ๊ทธ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์ด์šฉํ•˜์—ฌ
09:11
to build our first few personalized voices.
212
551721
4472
๋ช‡๋ช‡ ๊ฐœ๋ณ„ํ™”๋œ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋งŒ๋“ค์–ด๋ƒˆ์–ด์š”.
09:16
But there's so much more work to be done.
213
556193
1756
ํ•˜์ง€๋งŒ ์•„์ง๋„ ํ•ด์•ผํ•  ์ผ์ด ๋งŽ์Šต๋‹ˆ๋‹ค.
09:17
For Samantha, her surrogate
214
557949
2188
์‚ฌ๋งŒ๋‹ค์˜ ๊ฒฝ์šฐ์—, ๊ทธ ์•„์ด๊ฐ€ ๋นŒ๋ฆฐ ๋ชฉ์†Œ๋ฆฌ๋Š”
09:20
came from somewhere in the Midwest, a stranger
215
560137
3046
์ค‘์„œ๋ถ€ ์–ด๋Š ์ง€์—ญ์—์„œ ๋ฐ›์€๊ฑด๋ฐ
09:23
who gave her the gift of voice.
216
563183
3841
์–ผ๊ตด๋„ ๋ชจ๋ฅด๋Š” ์‚ฌ๋žŒ์ด ์„ ๋ฌผ์„ ์ค€ ๊ฒƒ์ด์ง€์š”.
09:27
And as a scientist, I'm so excited
217
567024
2153
๊ณผํ•™์ž๋กœ์„œ ์ €๋Š”
09:29
to take this work out of the laboratory
218
569177
1935
์‹คํ—˜์‹ค์—์„œ ์ด๋Ÿฐ ์ผ์„ ํ•ด๋‚ด๊ณ  ์‹ค์ œ ์„ธ์ƒ์—๋„ ํ™•์‚ฐํ•  ์ˆ˜ ์žˆ๋‹ค๋Š” ๊ฒƒ์—
09:31
and finally into the real world
219
571112
1800
์ƒ๋‹นํžˆ ๊ณ ๋ฌด๋˜์–ด ์žˆ์Šต๋‹ˆ๋‹ค.
09:32
so it can have real-world impact.
220
572912
3165
์‹ค์ƒํ™œ์— ์˜ํ–ฅ์„ ๋ฏธ์น  ์ˆ˜ ์žˆ๋Š” ๊ฒƒ์ด์ฃ .
09:36
What I want to share with you next
221
576077
1582
๊ทธ ๋‹ค์Œ์œผ๋กœ ์—ฌ๋Ÿฌ๋ถ„๋“ค๊ป˜
09:37
is how I envision taking this work
222
577659
2175
์ด๋Ÿฐ ์ž‘์—…์„ ์–ด๋–ป๊ฒŒ ๋‹ค์Œ ๋‹จ๊ณ„๋กœ
09:39
to that next level.
223
579834
2711
์ด๋Œ์–ด ๊ฐˆ์ง€์— ๋Œ€ํ•ด ๋ง์”€๋“œ๋ฆฌ๊ณ  ์‹ถ์–ด์š”.
09:42
I imagine a whole world of surrogate donors
224
582545
3887
์ €๋Š” ๋ชฉ์†Œ๋ฆฌ ๊ธฐ์ฆ์ž๊ฐ€ ์ „์„ธ๊ณ„์ ์œผ๋กœ ์ถฉ๋ถ„ํžˆ ์žˆ์–ด์„œ
09:46
from all walks of life, different sizes, different ages,
225
586432
3260
์–ด๋””๋ฅผ ๊ฐ€๋“ , ๋‹ค๋ฅธ ํฌ๊ธฐ์™€ ๋‚˜์ด๋Œ€์—์„œ ์žˆ๊ธฐ๋ฅผ ํฌ๋งํ•ฉ๋‹ˆ๋‹ค.
09:49
coming together in this voice drive
226
589692
3058
๋ชจ๋‘๊ฐ€ ์ด๋Ÿฐ ๋ชฉ์†Œ๋ฆฌ ๊ธฐ์ฆ์— ์ฐธ์—ฌํ•˜์—ฌ
09:52
to give people voices
227
592750
2270
์‚ฌ๋žŒ๋“ค์—๊ฒŒ
09:55
that are as colorful as their personalities.
228
595020
3799
๊ฐ์ž์˜ ๊ฐœ์„ฑ๋งŒํผ์ด๋‚˜ ๋‹ค์–‘ํ•œ ์Œ์ƒ‰์„ ์ค„ ์ˆ˜ ์žˆ๊ธฐ๋ฅผ ๋ฐ”๋ž๋‹ˆ๋‹ค.
09:58
To do that as a first step,
229
598819
2300
๊ทธ๋Ÿฌ๊ธฐ ์œ„ํ•œ ์ฒซ๊ฑธ์Œ์œผ๋กœ
10:01
we've put together this website, VocaliD.org,
230
601119
3275
์ €ํฌ๊ฐ€ ์ด๋Ÿฐ ์›น์‚ฌ์ดํŠธ๋ฅผ ๊ฐœ์„คํ–ˆ์–ด์š”. VocaliD.org ์ž…๋‹ˆ๋‹ค.
10:04
as a way to bring together those
231
604394
1624
์ด๋ฅผ ํ†ตํ•˜์—ฌ ๋ชฉ์†Œ๋ฆฌ ๊ธฐ์ฆ์ž๋กœ
10:06
who want to join us as voice donors,
232
606018
2675
์šฐ๋ฆฌ์™€ ํ•จ๊ป˜ ํ•˜๊ณ ์ž ํ•˜๋Š” ์‚ฌ๋žŒ๋“ค์„ ํ•œ๋ฐ ๋ชจ์œผ๊ณ 
10:08
as expertise donors,
233
608693
1772
๊ธฐ์ฆ๋„ ์ „๋ฌธํ™”ํ•˜๋Š” ๊ฒ๋‹ˆ๋‹ค.
10:10
in whatever way to make this vision a reality.
234
610465
5339
๊ทธ๋ฆฌ๊ณ  ์–ด๋–ค ๋ฐฉ์‹์œผ๋กœ๋“  ์ด๋Ÿฐ ํฌ๋ง์„ ํ˜„์‹ค๋กœ ๋งŒ๋“ค๊ณ ์ž ํ•ฉ๋‹ˆ๋‹ค.
10:15
They say that giving blood can save lives.
235
615804
4153
์‚ฌ๋žŒ๋“ค์€ ํ—Œํ˜ˆ์ด ์ƒ๋ช…์„ ๊ตฌํ•  ์ˆ˜ ์žˆ๋‹ค๊ณ  ๋งํ•ฉ๋‹ˆ๋‹ค.
10:19
Well, giving your voice can change lives.
236
619957
4982
์ž, ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ธฐ์ฆํ•˜๋ฉด ๋ˆ„๊ตฐ๊ฐ€์˜ ์‚ถ์„ ๋ฐ”๊ฟ€ ์ˆ˜ ์ž‡์–ด์š”.
10:24
All we need is a few hours of speech
237
624939
3050
ํ•ด์•ผํ•  ๊ฒƒ์ด๋ผ๊ณค ๋ชฉ์†Œ๋ฆฌ ๊ธฐ์ฆ์ž๋กœ
10:27
from our surrogate talker,
238
627989
1491
๋ช‡ ์‹œ๊ฐ„ ๋ฐœ์„ฑํ•˜๋Š” ๊ฒƒ์ด ์ „๋ถ€์—์š”.
10:29
and as little as a vowel from our target talker,
239
629480
4733
๋ชฉ์†Œ๋ฆฌ๊ฐ€ ํ•„์š”ํ•œ ์‚ฌ๋žŒ์ด ๋‚ด๋Š” ๋ช‡ ๊ฐ€์ง€ ๋ชจ์Œ์œผ๋กœ
10:34
to create a unique vocal identity.
240
634213
3711
๊ฐ์ž์˜ ๊ฐœ์„ฑ์— ๋งž๋Š” ์Œ์„ฑ์„ ์ฐฝ์กฐํ•ด๋‚ด๋Š” ๊ฒ๋‹ˆ๋‹ค.
10:37
So that's the science behind what we're doing.
241
637924
2626
๊ทธ๊ฒŒ ๋ฐ”๋กœ ์ €ํฌ๊ฐ€ ํ•˜๊ณ  ์žˆ๋Š” ์ผ์˜ ๋ฐ‘๋ฐ”ํƒ•์— ๊น”๋ฆฐ ๊ณผํ•™์ž…๋‹ˆ๋‹ค.
10:40
I want to end by circling back to the human side
242
640550
4455
์ €๋Š” ์ธ๊ฐ„์ ์ธ ์ธก๋ฉด์œผ๋กœ ๋˜๋Œ์•„ ๊ฐ€๋ฉด์„œ ๊ฐ•์—ฐ์„ ๋งˆ์น˜๊ณ ์ž ํ•ฉ๋‹ˆ๋‹ค.
10:45
that is really the inspiration for this work.
243
645005
4102
์ด๋Ÿฐ ์ž‘์—…์„ ํ•˜๊ฒŒ ๋œ ์ด์œ ์ด๊ธฐ๋„ ํ•˜์ฃ .
10:49
About five years ago, we built our very first voice
244
649107
3699
5๋…„์ „ ์ฆˆ์Œ, ์ €ํฌ๋Š” ์œŒ๋ฆฌ์—„์ด๋ž€ ์•„์ด๋ฅผ ์œ„ํ•ด
10:52
for a little boy named William.
245
652806
2501
์ฒ˜์Œ์œผ๋กœ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋งŒ๋“ค์—ˆ์—ˆ์–ด์š”.
10:55
When his mom first heard this voice,
246
655307
2357
์•„์ด์˜ ์—„๋งˆ๊ฐ€ ๊ทธ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์ฒ˜์Œ ๋“ค์—ˆ์„ ๋•Œ
10:57
she said, "This is what William
247
657664
2345
๊ทธ๋…€๊ฐ€ ์ด๋ ‡๊ฒŒ ๋งํ–ˆ์Šต๋‹ˆ๋‹ค.
11:00
would have sounded like
248
660009
1546
"์ด ์Œ์„ฑ์ด ์œŒ๋ฆฌ์—„์ด ๋ง์„ ํ•  ์ˆ˜ ์žˆ๋‹ค๋ฉด
11:01
had he been able to speak."
249
661555
2449
๋‚ด๋ฆฌ๋ผ๊ณ  ์ƒ๊ฐํ•˜๋Š” ์Œ์„ฑ์ด์—์š”."
11:04
And then I saw William typing a message
250
664004
2418
๊ทธ๋ฆฌ๊ณ ๋Š” ์œŒ๋ฆฌ์—„์— ์ œ๊ฒŒ ์ž์‹ ์˜ ๊ธฐ๊ตฌ๋ฅผ ํ†ตํ•ด
11:06
on his device.
251
666422
1362
๋ฌธ์ž ๋ฉ”์‹œ์ง€๋ฅผ ์“ฐ๋Š” ๊ฒƒ์„ ๋ณด์•˜์Šต๋‹ˆ๋‹ค.
11:07
I wondered, what was he thinking?
252
667784
3293
์ €๋„ ๊ถ์Œํ–ˆ์–ด์š”. ์•„์ด๊ฐ€ ์–ด๋–ป๊ฒŒ ์ƒ๊ฐํ• ๊นŒ?
11:11
Imagine carrying around someone else's voice
253
671077
3590
๋‚จ์˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๊ฐ€์ง€๊ณ  9๋…„์ด๋‚˜ ์‚ด์•„์˜ค๋‹ค๊ฐ€
11:14
for nine years
254
674667
2193
์ด์ œ์„œ์•ผ
11:16
and finally finding your own voice.
255
676860
4844
์ž๊ธฐ ์ž์‹ ๋งŒ์˜ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ์ฐพ๊ฒŒ ๋œ ๊ฒƒ์„ ์ƒ๊ฐํ•ด๋ณด์„ธ์š”.
11:21
Imagine that.
256
681704
1377
๊ทธ๊ฒŒ ์–ด๋–ค ๊ธฐ๋ถ„์ผ์ง€ ํ•œ๋ฒˆ ์ƒ์ƒํ•ด ๋ณด์„ธ์š”.
11:23
This is what William said:
257
683081
2797
์œŒ๋ฆฌ์—„์— ์ œ๊ฒŒ ํ•œ ๋ง์ž…๋‹ˆ๋‹ค:
11:25
"Never heard me before."
258
685878
4463
"์ด์ œ์•ผ ์ œ ๋ชฉ์†Œ๋ฆฌ๋ฅผ ๋“ค์–ด๋ณด๋„ค์š”."
11:32
Thank you.
259
692417
1619
๊ฐ์‚ฌํ•ฉ๋‹ˆ๋‹ค.
11:34
(Applause)
260
694036
4724
(๋ฐ•์ˆ˜)
์ด ์›น์‚ฌ์ดํŠธ ์ •๋ณด

์ด ์‚ฌ์ดํŠธ๋Š” ์˜์–ด ํ•™์Šต์— ์œ ์šฉํ•œ YouTube ๋™์˜์ƒ์„ ์†Œ๊ฐœํ•ฉ๋‹ˆ๋‹ค. ์ „ ์„ธ๊ณ„ ์ตœ๊ณ ์˜ ์„ ์ƒ๋‹˜๋“ค์ด ๊ฐ€๋ฅด์น˜๋Š” ์˜์–ด ์ˆ˜์—…์„ ๋ณด๊ฒŒ ๋  ๊ฒƒ์ž…๋‹ˆ๋‹ค. ๊ฐ ๋™์˜์ƒ ํŽ˜์ด์ง€์— ํ‘œ์‹œ๋˜๋Š” ์˜์–ด ์ž๋ง‰์„ ๋”๋ธ” ํด๋ฆญํ•˜๋ฉด ๊ทธ๊ณณ์—์„œ ๋™์˜์ƒ์ด ์žฌ์ƒ๋ฉ๋‹ˆ๋‹ค. ๋น„๋””์˜ค ์žฌ์ƒ์— ๋งž์ถฐ ์ž๋ง‰์ด ์Šคํฌ๋กค๋ฉ๋‹ˆ๋‹ค. ์˜๊ฒฌ์ด๋‚˜ ์š”์ฒญ์ด ์žˆ๋Š” ๊ฒฝ์šฐ ์ด ๋ฌธ์˜ ์–‘์‹์„ ์‚ฌ์šฉํ•˜์—ฌ ๋ฌธ์˜ํ•˜์‹ญ์‹œ์˜ค.

https://forms.gle/WvT1wiN1qDtmnspy7