ABOUT THE SPEAKER
Michael Rubinstein - Research scientist, Google
Computer scientist Michael Rubinstein and his team have developed a "motion microscope" that can show video footage of barely perceivable movements, like breaths and heartbeats.

Why you should listen

Michael Rubinstein zooms in on what we can't see and mangnifies it by thirty or a hundred times. His "motion microscope," developed at MIT with Microsoft and Quanta Research, picks up on subtle motion and color changes in videos and blows them up for the naked eye to see. The result: fun, cool, creepy videos.

Rubinstein is a research scientist at a new Cambridge-based Google lab for computer vision research. He has a PhD in computer science and electrical engineering from MIT.

More profile about the speaker
Michael Rubinstein | Speaker | TED.com
TEDxBeaconStreet

Michael Rubinstein: See invisible motion, hear silent sounds

麥可·魯賓斯坦: 看到無法看見的動作,聽見無法聽到的聲音。酷炫?詭異?我們尚不能定論

Filmed:
2,075,056 views

「動作觀察顯微鏡」是一種影像處理工具,能放大肉眼無法觀察的聲音、顏色變化讓我們看見。影像研究者麥可·魯賓斯坦令人瞠目結舌的影片中,展示透過簡單的鏡頭如何能夠紀錄脈搏和心跳。觀看他如何透過放大洋芋片包裝袋因聲波產生的震動來還原一段對話。一個驚世、邪異之作你必須眼見為憑。
- Research scientist, Google
Computer scientist Michael Rubinstein and his team have developed a "motion microscope" that can show video footage of barely perceivable movements, like breaths and heartbeats. Full bio

Double-click the English transcript below to play the video.

00:13
So over the past過去 few少數 centuries百年,
microscopes顯微鏡 have revolutionized革命性 our world世界.
0
1114
7492
幾世紀以來,
顯微鏡對世界產生革命性的影響
00:21
They revealed透露 to us a tiny world世界
of objects對象, life and structures結構
1
9036
5216
它為我們揭開了微小世界的
物體、生命和結構
00:26
that are too small for us
to see with our naked eyes眼睛.
2
14252
2906
那些因為太過微小而沒有辦法用肉眼觀察
00:29
They are a tremendous巨大 contribution貢獻
to science科學 and technology技術.
3
17158
3019
對於科學和科技是個巨大的貢獻
00:32
Today今天 I'd like to introduce介紹 you
to a new type類型 of microscope顯微鏡,
4
20177
3227
今天我想介紹大家一種新的顯微鏡
00:35
a microscope顯微鏡 for changes變化.
5
23404
2578
一種觀察變化的顯微鏡
00:37
It doesn't use optics光學
like a regular定期 microscope顯微鏡
6
25982
2902
不像一般顯微鏡使用光線
00:40
to make small objects對象 bigger,
7
28884
1997
放大微小的物體
00:42
but instead代替 it uses使用 a video視頻 camera相機
and image圖片 processing處理
8
30881
4376
而是使用攝影機和圖像處理
00:47
to reveal揭示 to us the tiniest最小的 motions運動
and color顏色 changes變化 in objects對象 and people,
9
35257
5256
來讓我們看見
人與物的微小運動和顏色變化
00:52
changes變化 that are impossible不可能
for us to see with our naked eyes眼睛.
10
40513
3842
那些變化是我們沒有辦法用肉眼觀察的
00:56
And it lets讓我們 us look at our world世界
in a completely全然 new way.
11
44355
4120
讓我們以全新的方式去看世界
01:00
So what do I mean by color顏色 changes變化?
12
48475
1910
而我所說的顏色變化到底是甚麼?
01:02
Our skin皮膚, for example,
changes變化 its color顏色 very slightly
13
50385
2832
舉例來說:皮膚
我們的皮膚會產生極細微的顏色變化
01:05
when the blood血液 flows流動 under it.
14
53217
1997
當我們的血液流過
01:07
That change更改 is incredibly令人難以置信 subtle微妙,
15
55214
2397
這個變化是極度細微的
01:09
which哪一個 is why, when you
look at other people,
16
57611
2063
這也是為什麼當大家看向別人時
01:11
when you look at the person
sitting坐在 next下一個 to you,
17
59674
2251
當你看著坐在你身旁的人
01:13
you don't see their skin皮膚
or their face面對 changing改變 color顏色.
18
61925
3575
你不會看到他臉部皮膚的顏色變化
01:17
When we look at this video視頻 of Steve史蒂夫 here,
it appears出現 to us like a static靜態的 picture圖片,
19
65500
4360
當我們看到影片中的史蒂夫,
顯示出的是一張靜態的圖片
01:21
but once一旦 we look at this video視頻
through通過 our new, special特別 microscope顯微鏡,
20
69860
3860
但當我們以新的、特殊的顯微鏡
01:25
suddenly突然 we see
a completely全然 different不同 image圖片.
21
73720
2600
突然間看見一個全然不同的影像
01:28
What you see here are small changes變化
in the color顏色 of Steve's史蒂夫的 skin皮膚,
22
76320
3930
這裡你們看見的是
史蒂夫臉部皮膚的微小顏色變化
放大一百倍後
01:32
magnified放大 100 times
so that they become成為 visible可見.
23
80250
4436
變得肉眼可見
我們可以真的看見
01:36
We can actually其實 see a human人的 pulse脈衝.
24
84686
3267
人類的脈搏
我們可以看見史蒂夫的心跳
是多麼的快速
01:39
We can see how fast快速
Steve's史蒂夫的 heart is beating跳動,
25
87953
3227
而且我們也看見
血液實際上如何流過他的臉部
01:43
but we can also see the actual實際 way
that the blood血液 flows流動 in his face面對.
26
91180
5355
01:48
And we can do that not just
to visualize想像 the pulse脈衝,
27
96544
2631
而且我們不只是將脈搏變得可見
01:51
but also to actually其實
recover恢復 our heart rates利率,
28
99175
3471
還能夠真的獲取我們的心跳
01:54
and measure測量 our heart rates利率.
29
102646
1793
和測量我們的心跳速率
01:56
And we can do it with regular定期 cameras相機
and without touching接觸 the patients耐心.
30
104439
4453
並且我們可以使用普通的鏡頭
而不需要觸碰病人
02:00
So here you see the pulse脈衝 and heart rate
we extracted提取 from a neonatal新生兒 baby寶寶
31
108892
5617
因此在這裡你們可以看見
得自新生嬰兒的脈搏和心跳速率
02:06
from a video視頻 we took
with a regular定期 DSLRDSLR camera相機,
32
114509
2881
透過我們攝自普通的數位單眼相機
02:09
and the heart rate measurement測量 we get
33
117390
1816
我們測量出心跳速率
02:11
is as accurate準確 as the one you'd get
with a standard標準 monitor監控 in a hospital醫院.
34
119206
4811
是與醫院中標準監測設備一樣準確的
02:16
And it doesn't even have to be
a video視頻 we recorded記錄.
35
124017
2642
而且甚至不需要使用我們拍攝的影片
02:18
We can do it essentially實質上
with other videos視頻 as well.
36
126659
2995
我們可以實質上使用在其他人拍攝的影片
02:21
So I just took a short clip
from "Batman蝙蝠俠 Begins開始" here
37
129654
3901
因此我擷取了
"蝙蝠俠:開戰時刻"裡的一個片段
02:25
just to show顯示 Christian基督教 Bale's貝爾的 pulse脈衝.
38
133555
1904
來顯示克里斯汀·貝爾的脈搏
02:27
(Laughter笑聲)
39
135459
1822
(大笑)
02:29
And you know, presumably想必
he's wearing穿著 makeup化妝,
40
137281
2123
而你們知道,他大概有化妝
02:31
the lighting燈光 here is kind of challenging具有挑戰性的,
41
139404
1953
這樣的明暗變化有點挑戰性
02:33
but still, just from the video視頻,
we're able能夠 to extract提取 his pulse脈衝
42
141357
2951
但還是一樣,在這個影片
我們仍能夠取得他的脈搏
02:36
and show顯示 it quite相當 well.
43
144308
2018
而且還蠻明顯的
02:38
So how do we do all that?
44
146326
1920
然而我們到底是怎麼做到的?
02:40
We basically基本上 analyze分析 the changes變化
in the light that are recorded記錄
45
148246
4598
基本上,我們分析記錄到的光線變化
02:44
at every一切 pixel像素 in the video視頻 over time,
46
152844
2271
每一個畫面中的像素
02:47
and then we crank曲柄 up those changes變化.
47
155115
1798
然後我們處理這些變化
02:48
We make them bigger
so that we can see them.
48
156913
2162
我們放大這些變化讓它們可以被看見
02:51
The tricky狡猾 part部分 is that those signals信號,
49
159075
1902
其中最難處理的是那些訊號
02:52
those changes變化 that we're after,
are extremely非常 subtle微妙,
50
160977
2933
那些我們想要尋找的變化,
是非常細微的
02:55
so we have to be very careful小心
when you try to separate分離 them
51
163910
2779
所以當我們分離出變化時必須非常細心
02:58
from noise噪聲 that always exists存在 in videos視頻.
52
166689
3831
因為干擾會一直出現在影像中
03:02
So we use some clever聰明
image圖片 processing處理 techniques技術
53
170520
2995
因此我們使用巧妙的圖像處理技巧
03:05
to get a very accurate準確 measurement測量
of the color顏色 at each pixel像素 in the video視頻,
54
173515
3994
自影像裡每個像素取得
非常精確的顏色辨識
03:09
and then the way the color顏色
changes變化 over time,
55
177509
2670
和隨著時間產生的顏色變化
03:12
and then we amplify放大 those changes變化.
56
180179
2693
然後我們放大這些變化
03:14
We make them bigger to create創建 those types類型
of enhanced增強 videos視頻, or magnified放大 videos視頻,
57
182872
3980
我們讓變化增大制出改進的影像
或是放大的影像
03:18
that actually其實 show顯示 us those changes變化.
58
186852
2172
而確實地向我們顯示出變化來
03:21
But it turns out we can do that
not just to show顯示 tiny changes變化 in color顏色,
59
189024
4238
[運動顯微鏡]
但後來我們不僅能展示微小的顏色變化
03:25
but also tiny motions運動,
60
193262
2241
還能表現出細微的動作
03:27
and that's because the light
that gets得到 recorded記錄 in our cameras相機
61
195503
3576
那是因為我們鏡頭紀錄的光線
03:31
will change更改 not only if the color顏色
of the object目的 changes變化,
62
199079
2810
不只有物體顏色上的變化
03:33
but also if the object目的 moves移動.
63
201889
2368
還有物體的運動
03:36
So this is my daughter女兒
when she was about two months個月 old.
64
204257
3636
這是我女兒,當時兩個月大
03:39
It's a video視頻 I recorded記錄
about three years年份 ago.
65
207893
2999
這是我三年前拍攝的影片
03:42
And as new parents父母, we all want
to make sure our babies嬰兒 are healthy健康,
66
210892
3208
而且像所有的新手父母親一樣,
我希望我們的小孩健康
03:46
that they're breathing呼吸,
that they're alive, of course課程.
67
214100
2542
他們有沒有呼吸,還活著,想當然
03:48
So I too got one of those baby寶寶 monitors顯示器
68
216642
2142
所以我也買了嬰兒監視器
03:50
so that I could see my daughter女兒
when she was asleep睡著.
69
218784
2469
讓我能夠在我女兒睡覺時看見她
03:53
And this is pretty漂亮 much what you'll你會 see
with a standard標準 baby寶寶 monitor監控.
70
221253
3527
一般嬰兒監視器拍到的
差不多就像這樣
03:56
You can see the baby's寶寶 sleeping睡眠, but
there's not too much information信息 there.
71
224780
3682
你知道的,你可以看到嬰兒的睡眠
但資訊顯現的不多
04:00
There's not too much we can see.
72
228474
1604
我們能看見的不多
04:02
Wouldn't豈不 it be better,
or more informative信息, or more useful有用,
73
230078
2824
如果換成這種觀看的視角
04:04
if instead代替 we could look
at the view視圖 like this.
74
232902
2990
更多的訊息、更有幫助會好很多
04:07
So here I took the motions運動
and I magnified放大 them 30 times,
75
235892
6356
所以我拍攝了這些動作並且放大30倍
04:14
and then I could clearly明確地 see that my
daughter女兒 was indeed確實 alive and breathing呼吸.
76
242248
3826
然後我可以清楚看見我的女兒
的確活著而且有呼吸
04:18
(Laughter笑聲)
77
246074
2253
(大笑)
04:20
Here is a side-by-side並排側 comparison對照.
78
248327
1922
這裡並列的是比較圖
04:22
So again, in the source資源 video視頻,
in the original原版的 video視頻,
79
250249
2483
在原始的影像裡
04:24
there's not too much we can see,
80
252732
1636
並沒有太多可以觀察的
04:26
but once一旦 we magnify放大 the motions運動,
the breathing呼吸 becomes much more visible可見.
81
254368
3707
而一旦經過我們放大動作,
呼吸變得更容易觀察
04:30
And it turns out, there's
a lot of phenomena現象
82
258075
2070
而這表示,許多現象的出現
04:32
we can reveal揭示 and magnify放大
with our new motion運動 microscope顯微鏡.
83
260145
3623
透過新式運動顯微鏡我們可以找且放大
04:35
We can see how our veins and arteries動脈
are pulsing脈動 in our bodies身體.
84
263768
4564
我們可以觀察我們的動脈和靜脈
如何在我們身體裡脈動
04:40
We can see that our eyes眼睛
are constantly經常 moving移動
85
268332
2628
可以看見我們的眼睛是不斷的移動
04:42
in this wobbly搖擺不定 motion運動.
86
270960
1887
不穩定的擺動
04:44
And that's actually其實 my eye,
87
272847
1509
並且那實際上是我的眼睛
04:46
and again this video視頻 was taken採取
right after my daughter女兒 was born天生,
88
274356
3065
再次,這個影片拍攝就在我女兒出生之後
04:49
so you can see I wasn't getting得到
too much sleep睡覺. (Laughter笑聲)
89
277421
4202
所以你看得出我那時候沒甚麼睡
(大笑)
04:53
Even when a person is sitting坐在 still,
90
281623
2716
就算一個人坐著
04:56
there's a lot of information信息
we can extract提取
91
284339
2044
還是有許多的訊息我們能夠獲取
04:58
about their breathing呼吸 patterns模式,
small facial面部 expressions表達式.
92
286383
3529
他們呼吸的模式、細微的臉部表情
05:01
Maybe we could use those motions運動
93
289912
1625
說不定我們可以運用這些運動
05:03
to tell us something about
our thoughts思念 or our emotions情緒.
94
291537
3154
從而得知我們的想法或情緒
05:06
We can also magnify放大 small
mechanical機械 movements運動,
95
294691
3255
我們可以放大細微的機械運動
05:09
like vibrations振動 in engines引擎,
96
297946
1555
像是引擎的顫動
05:11
that can help engineers工程師 detect檢測
and diagnose診斷 machinery機械 problems問題,
97
299501
3692
這可以幫助工程師檢測和診斷出機械問題
05:15
or see how our buildings房屋 and structures結構
sway搖擺 in the wind and react應對 to forces軍隊.
98
303193
4738
或是我們的大樓和建築物
在風吹、地震影響下的晃動
05:19
Those are all things that our society社會
knows知道 how to measure測量 in various各個 ways方法,
99
307931
4581
在這個社會中知道
各種不同測量物體的方式
05:24
but measuring測量 those motions運動 is one thing,
100
312512
2453
但這種測量是另一回事
05:26
and actually其實 seeing眼看 those
motions運動 as they happen發生
101
314965
2276
真的看見這些運動的發生
05:29
is a whole整個 different不同 thing.
102
317241
2554
是完全不同的事情
05:31
And ever since以來 we discovered發現
this new technology技術,
103
319795
3041
自我們發現這個新科技後
05:34
we made製作 our code available可得到 online線上 so that
others其他 could use and experiment實驗 with it.
104
322836
3953
我們就將編碼放在網路上
讓其他人能夠使用並測試它
05:38
It's very simple簡單 to use.
105
326789
1875
它是非常容易上手的
05:40
It can work on your own擁有 videos視頻.
106
328664
2044
它能夠運用在你的影片上
05:42
Our collaborators合作者 at Quanta廣達 Research研究
even created創建 this nice不錯 website網站
107
330708
3193
我們量子研究的合作人
還架設了一個很好的網站
05:45
where you can upload上載 your videos視頻
and process處理 them online線上,
108
333901
2678
可以讓你上傳影片和在網路上進行處理
05:48
so even if you don't have any experience經驗
in computer電腦 science科學 or programming程序設計,
109
336579
3816
所以即使你沒有任何電腦科學
和寫軟體的經驗
05:52
you can still very easily容易 experiment實驗
with this new microscope顯微鏡.
110
340395
2936
還是很容易就能測試新式顯微鏡
05:55
And I'd like to show顯示 you
just a couple一對 of examples例子
111
343331
2404
並且在這裡我很樂意秀幾個例子給你們
05:57
of what others其他 have doneDONE with it.
112
345735
2735
其他人如何去使用它
06:00
So this video視頻 was made製作 by
a YouTubeYouTube的 user用戶 called TamezTamez85.
113
348470
5317
而這是 YouTube 使用者
Tamez85 所做的影片
06:05
I don't know who that user用戶 is,
114
353787
1463
我並不知道使用者是誰
06:07
but he, or she, used our code
115
355250
2345
但他或她,使用了編碼
06:09
to magnify放大 small belly肚皮
movements運動 during pregnancy懷孕.
116
357595
3715
放大懷孕中腹部的細微運動
06:13
It's kind of creepy爬行.
117
361310
1602
是有點詭異
06:14
(Laughter笑聲)
118
362912
1613
(大笑)
06:16
People have used it to magnify放大
pulsing脈動 veins in their hands.
119
364525
4961
有人使用它去放大手部的靜脈脈博
06:21
And you know it's not real真實 science科學
unless除非 you use guinea幾內亞 pigs,
120
369486
3782
你們也知道只有用天竺鼠測試過
才叫真的科學
06:25
and apparently顯然地 this guinea幾內亞 pig
is called Tiffany蒂芙尼,
121
373268
3390
這隻天竺鼠叫做蒂芬妮
06:28
and this YouTubeYouTube的 user用戶 claims索賠
it is the first rodent囓齒動物 on Earth地球
122
376658
2949
這位YouTube使用者表示
這是世界上第一隻囓齒動物
06:31
that was motion-magnified運動放大.
123
379607
2688
使用動作放大化
06:34
You can also do some art藝術 with it.
124
382295
2188
你還可以加入藝術
06:36
So this video視頻 was sent發送 to me
by a design設計 student學生 at Yale耶魯.
125
384483
3018
這個影片是一位
耶魯大學設計系的學生寄給我的
06:39
She wanted to see
if there's any difference區別
126
387501
2137
她想要看到
她同學的移動方式是否有所差異
06:41
in the way her classmates同學 move移動.
127
389638
1522
她讓同學們都站直不動
然後放大他們的動作
06:43
She made製作 them all stand still,
and then magnified放大 their motions運動.
128
391160
4209
06:47
It's like seeing眼看
still pictures圖片 come to life.
129
395369
3378
像把靜止的相片變得有生命
06:50
And the nice不錯 thing with
all those examples例子
130
398747
2433
而這些不錯的例子
06:53
is that we had nothing to do with them.
131
401180
2296
完全不是我們做出的
06:55
We just provided提供 this new tool工具,
a new way to look at the world世界,
132
403476
3854
我們只提供新工具,
一種新的方式去觀察世界
06:59
and then people find other interesting有趣,
new and creative創作的 ways方法 of using運用 it.
133
407330
5132
然後有人發現一些
有趣、有創意的新方式去使用
07:04
But we didn't stop there.
134
412462
1764
而我們沒有就停在這裡
07:06
This tool工具 not only allows允許 us
to look at the world世界 in a new way,
135
414226
3251
這個工具並非只讓我們用新的方式看世界
07:09
it also redefines重新定義 what we can do
136
417477
2368
它還重新定義我們能夠做的
07:11
and pushes the limits範圍 of what
we can do with our cameras相機.
137
419845
3181
並且將相機的使用推向極限
07:15
So as scientists科學家們, we started開始 wondering想知道,
138
423026
2229
所以身為科學家,我們開始想
07:17
what other types類型 of physical物理 phenomena現象
produce生產 tiny motions運動
139
425255
3785
有哪些物理現象會產生細微的運動
07:21
that we could now use
our cameras相機 to measure測量?
140
429040
2903
可以讓我們用相機去測量
07:23
And one such這樣 phenomenon現象
that we focused重點 on recently最近 is sound聲音.
141
431943
4001
最近我們專注於聲音這個現象
07:27
Sound聲音, as we all know,
is basically基本上 changes變化
142
435944
2105
聲音就我們所知是一種
07:30
in air空氣 pressure壓力 that
travel旅行 through通過 the air空氣.
143
438049
2183
大氣壓力的變化通過空氣
07:32
Those pressure壓力 waves波浪 hit擊中 objects對象
and they create創建 small vibrations振動 in them,
144
440232
3621
壓力波動打在物體上使其產生微小的顫動
07:35
which哪一個 is how we hear
and how we record記錄 sound聲音.
145
443853
2532
而我們透過它聽見和紀錄聲音
07:38
But it turns out that sound聲音
also produces產生 visual視覺 motions運動.
146
446385
3668
結果我們還發現聲音會產生可見的運動
07:42
Those are motions運動
that are not visible可見 to us
147
450053
2833
那些不能用肉眼觀察
07:44
but are visible可見 to a camera相機
with the right processing處理.
148
452886
3001
但相機可以看見且處理的
07:47
So here are two examples例子.
149
455887
1573
而這裡是兩個例子
07:49
This is me demonstrating示範
my great singing唱歌 skills技能.
150
457460
2614
這裡展示著我厲害的歌唱能力
07:53
(Singing唱歌)
151
461064
1634
(唱歌)
07:54
(Laughter笑聲)
152
462698
1436
(大笑)
07:56
And I took a high-speed高速 video視頻
of my throat while I was humming低唱.
153
464134
2986
而我拍攝了一段當我在唱歌時
我喉嚨的高速影片
07:59
Again, if you stare at that video視頻,
154
467120
1764
再次,如果你注意這個影片
08:00
there's not too much
you'll你會 be able能夠 to see,
155
468884
2076
並沒有太多東西你能看到的
但一旦以 100 倍放大那些動作
08:02
but once一旦 we magnify放大 the motions運動 100 times,
we can see all the motions運動 and ripples漣漪
156
470960
4332
我們可以看見運動和波紋
08:07
in the neck頸部 that are involved參與
in producing生產 the sound聲音.
157
475292
3274
在脖子裡面,參與聲音的製造
08:10
That signal信號 is there in that video視頻.
158
478566
2740
這個聲音訊息就在這部影片中
08:13
We also know that singers歌手
can break打破 a wine紅酒 glass玻璃
159
481306
2670
我們還知道歌手可以唱破酒杯
08:15
if they hit擊中 the correct正確 note注意.
160
483976
1463
假設他們抓到了對的音
08:17
So here, we're going to play a note注意
161
485439
1765
所以這裡我們要來播放一個音
08:19
that's in the resonance諧振
frequency頻率 of that glass玻璃
162
487204
2526
在玻璃的共振頻率範圍內
08:21
through通過 a loudspeaker喇叭 that's next下一個 to it.
163
489730
2048
透過旁邊的擴音器
08:23
Once一旦 we play that note注意
and magnify放大 the motions運動 250 times,
164
491778
4419
一旦我們放出並且放大動作 250 倍
08:28
we can very clearly明確地 see
how the glass玻璃 vibrates振動
165
496197
2338
可以清晰地看到玻璃是如何顫動
08:30
and resonates共振 in response響應 to the sound聲音.
166
498535
3570
和與聲音的共振
08:34
It's not something you're used
to seeing眼看 every一切 day.
167
502105
2420
這並非我們平日能夠所見
08:36
But this made製作 us think.
It gave us this crazy idea理念.
168
504525
3529
但這讓我們思考。
給我們一個瘋狂的想法
08:40
Can we actually其實 invert倒置 this process處理
and recover恢復 sound聲音 from video視頻
169
508054
5608
我們能夠反過來處理
並恢復影像中的聲音嗎?
08:45
by analyzing分析 the tiny vibrations振動
that sound聲音 waves波浪 create創建 in objects對象,
170
513662
4035
透過分析這些由聲音
影響物體產生的微小顫動
08:49
and essentially實質上 convert兌換 those
back into the sounds聲音 that produced生成 them.
171
517697
4777
然後實際上還原出源頭的聲音
08:54
In this way, we can turn
everyday每天 objects對象 into microphones麥克風.
172
522474
4457
透過這個方式,我們可以
讓所有東西都變成麥克風
08:58
So that's exactly究竟 what we did.
173
526931
2232
而這裡是我們實際做的
09:01
So here's這裡的 an empty bag of chips芯片
that was lying說謊 on a table,
174
529163
2816
一個空的洋芋片包裝放在桌上
09:03
and we're going to turn that
bag of chips芯片 into a microphone麥克風
175
531979
2825
然後我們要將那包洋芋片
藉由拍攝將它變成我們的麥克風
09:06
by filming拍戲 it with a video視頻 camera相機
176
534804
1591
並且分析它因為聲音影響
而生的細微運動
09:08
and analyzing分析 the tiny motions運動
that sound聲音 waves波浪 create創建 in it.
177
536395
3228
09:11
So here's這裡的 the sound聲音
that we played發揮 in the room房間.
178
539623
2796
這是我們在房間裡撥放的聲音
09:14
(Music音樂: "Mary瑪麗 Had a Little Lamb羊肉")
179
542419
4434
(音樂:Mary Had a Little Lamb)
09:22
And this is a high-speed高速 video視頻
we recorded記錄 of that bag of chips芯片.
180
550007
3025
這裡是我們拍攝包裝袋的高速影像
09:25
Again it's playing播放.
181
553032
1274
聲音在撥放
09:26
There's no chance機會 you'll你會 be able能夠
to see anything going on in that video視頻
182
554306
3342
若只是用看的
你無法看見影片中發生甚麼事
09:29
just by looking at it,
183
557648
1058
但這裡我們能還原聲音
09:30
but here's這裡的 the sound聲音 we were able能夠
to recover恢復 just by analyzing分析
184
558706
2984
就只依靠分析影片中的細微運動
09:33
the tiny motions運動 in that video視頻.
185
561690
2183
09:35
(Music音樂: "Mary瑪麗 Had a Little Lamb羊肉")
186
563873
2809
(音樂:Mary Had a Little Lamb)
09:52
I call it -- Thank you.
187
580985
1486
我稱呼它 -- 謝謝
09:54
(Applause掌聲)
188
582471
5225
(掌聲)
10:01
I call it the visual視覺 microphone麥克風.
189
589878
2345
我稱呼它為視覺麥克風
10:04
We actually其實 extract提取 audio音頻 signals信號
from video視頻 signals信號.
190
592223
3390
我們真的從影像訊號中擷取聲音訊號
10:07
And just to give you a sense
of the scale規模 of the motions運動 here,
191
595613
3181
而這裡給你了解運動的規模
10:10
a pretty漂亮 loud sound聲音 will cause原因 that bag
of chips芯片 to move移動 less than a micrometer千分尺.
192
598799
5336
響亮的聲音只會讓
洋芋片包裝袋移動不到一微米
10:16
That's one thousandth千分之一 of a millimeter毫米.
193
604135
2739
那是一千分之一毫米
10:18
That's how tiny the motions運動 are
that we are now able能夠 to pull out
194
606874
3561
這就是我們現在找出的動作,如此細微
10:22
just by observing觀察 how light
bounces反彈 off objects對象
195
610435
3243
只透過反射自物體的光線
10:25
and gets得到 recorded記錄 by our cameras相機.
196
613678
2136
並且用我們的相機記錄
10:27
We can recover恢復 sounds聲音
from other objects對象, like plants植物.
197
615814
3250
我們可以從物體上還原聲音,像植物
10:31
(Music音樂: "Mary瑪麗 Had a Little Lamb羊肉")
198
619064
6316
(音樂:Mary Had a Little Lamb)
10:39
And we can recover恢復 speech言語 as well.
199
627214
1997
還可以還原言語
10:41
So here's這裡的 a person speaking請講 in a room房間.
200
629211
2577
使用喇叭,這是一個人在房間裡講話
10:43
Voice語音: Mary瑪麗 had a little lamb羊肉
whose誰的 fleece羊毛 was white白色 as snow,
201
631788
4203
聲音:瑪莉有一隻小羊,
牠的毛和雪一樣白
10:47
and everywhere到處 that Mary瑪麗 went,
that lamb羊肉 was sure to go.
202
635991
4230
而不論瑪莉去哪裡,都會帶著小羊
10:52
Michael邁克爾 Rubinstein魯賓斯坦: And here's這裡的
that speech言語 again recovered恢復
203
640221
2759
麥可·魯賓斯坦:
這裡是那段話被再次還原
10:54
just from this video視頻
of that same相同 bag of chips芯片.
204
642980
3274
正是從這影片中同樣的洋芋片包裝袋
10:58
Voice語音: Mary瑪麗 had a little lamb羊肉
whose誰的 fleece羊毛 was white白色 as snow,
205
646254
4831
聲音:瑪莉有一隻小羊,
牠的毛和雪一樣白
11:03
and everywhere到處 that Mary瑪麗 went,
that lamb羊肉 was sure to go.
206
651085
4859
而不論瑪莉去哪裡,都會帶著小羊
11:07
MR先生: We used "Mary瑪麗 Had a Little Lamb羊肉"
207
655944
2346
我們使用
"Mary Had a Little Lamb"
11:10
because those are said to be
the first words
208
658290
2123
是因為這據說是愛迪生
11:12
that Thomas托馬斯 Edison愛迪生 spoke
into his phonograph留聲機 in 1877.
209
660413
4161
在 1877 年,第一次對他的留聲機所留
11:16
It was one of the first sound聲音
recording記錄 devices設備 in history歷史.
210
664574
3228
這是歷史上第一個聲音紀錄裝置
11:19
It basically基本上 directed針對 the sounds聲音
onto a diaphragm光圈
211
667802
3327
它基本上在一個隔膜上接受聲音
11:23
that vibrated振動 a needle that essentially實質上
engraved the sound聲音 on tinfoil錫紙
212
671129
4079
而隔膜震動了針而雕刻錫板
11:27
that was wrapped包裹 around the cylinder圓筒.
213
675208
2275
錫板是包裹在圓柱體上
11:29
Here's這裡的 a demonstration示範 of recording記錄 and
replaying重播 sound聲音 with Edison's愛迪生 phonograph留聲機.
214
677483
5943
此處展示的是愛迪生的留聲機
如何錄入和重放
11:35
(Video視頻) Voice語音: Testing測試,
testing測試, one two three.
215
683426
3020
(影片)聲音:測試,測試,一二三
11:38
Mary瑪麗 had a little lamb羊肉
whose誰的 fleece羊毛 was white白色 as snow,
216
686446
3413
瑪莉有隻小羊
羊毛像雪一般白
11:41
and everywhere到處 that Mary瑪麗 went,
the lamb羊肉 was sure to go.
217
689859
3669
而且瑪莉到哪裡,都會帶著小羊
11:45
Testing測試, testing測試, one two three.
218
693528
2740
測試,測試,一二三
11:48
Mary瑪麗 had a little lamb羊肉
whose誰的 fleece羊毛 was white白色 as snow,
219
696268
4156
瑪莉有隻小羊
羊毛像雪一般白
11:52
and everywhere到處 that Mary瑪麗 went,
the lamb羊肉 was sure to go.
220
700424
5224
而且瑪莉到哪裡,都會帶著小羊
11:57
MR先生: And now, 137 years年份 later後來,
221
705648
4017
而現在,137 年後的現在
12:01
we're able能夠 to get sound聲音
in pretty漂亮 much similar類似 quality質量
222
709665
4087
我們能夠擷取聲音到非常相似的程度
12:05
but by just watching觀看 objects對象
vibrate顫動 to sound聲音 with cameras相機,
223
713752
4079
卻只是藉由相機觀察物體對聲音的顫動
12:09
and we can even do that when the camera相機
224
717831
1934
我們還能做到這件事在相機
12:11
is 15 feet away from the object目的,
behind背後 soundproof隔音 glass玻璃.
225
719765
4234
離物體 15 英呎
在隔音玻璃外
而這是我們在這個情況下得到的聲音
12:15
So this is the sound聲音 that we were
able能夠 to recover恢復 in that case案件.
226
723999
3220
12:19
Voice語音: Mary瑪麗 had a little lamb羊肉
whose誰的 fleece羊毛 was white白色 as snow,
227
727219
5294
聲音:瑪莉有隻小羊
羊毛像雪一般白
12:24
and everywhere到處 that Mary瑪麗 went,
the lamb羊肉 was sure to go.
228
732513
4759
而且瑪莉到哪裡,都會帶著小羊
12:29
MR先生: And of course課程, surveillance監控 is
the first application應用 that comes to mind心神.
229
737404
3630
想當然,監控裝置
是第一個被想到的應用方式
12:33
(Laughter笑聲)
230
741034
2995
(大笑)
12:36
But it might威力 actually其實 be useful有用
for other things as well.
231
744029
4056
但這可能真的對其他事情很有用
12:40
Maybe in the future未來, we'll be able能夠
to use it, for example,
232
748085
2840
可能在未來我們能夠使用它,舉例來說
12:42
to recover恢復 sound聲音 across橫過 space空間,
233
750925
2252
在外太空還原聲音
12:45
because sound聲音 can't travel旅行
in space空間, but light can.
234
753177
3576
因為聲音不能在太空中傳播而光線可以
12:48
We've我們已經 only just begun開始 exploring探索
235
756753
2404
我們才剛開始探索
12:51
other possible可能 uses使用
for this new technology技術.
236
759157
3019
這個裝置其他的運用可能
12:54
It lets讓我們 us see physical物理 processes流程
that we know are there
237
762176
2832
它讓看見我們實際知道存在的物理運作
12:57
but that we've我們已經 never been able能夠
to see with our own擁有 eyes眼睛 until直到 now.
238
765008
3556
但直到現在我們才能用肉眼觀察
13:00
This is our team球隊.
239
768564
1204
這是我的團隊
13:01
Everything I showed顯示 you today今天
is a result結果 of a collaboration合作
240
769768
2879
每一個在今天展示出的東西是團隊成果
13:04
with this great group
of people you see here,
241
772647
2191
和你在這裡看到的一群偉大的人
13:06
and I encourage鼓勵 you and welcome歡迎 you
to check out our website網站,
242
774838
3167
而我鼓勵各位並且
歡迎各位看看我們的網站
13:10
try it out yourself你自己,
243
778005
1446
自己試試看
13:11
and join加入 us in exploring探索
this world世界 of tiny motions運動.
244
779451
2972
並且加入我們一起探索世界中的微小運動
13:14
Thank you.
245
782423
1625
謝謝大家
13:16
(Applause掌聲)
246
784048
1254
(掌聲)
Translated by HAO KUN SANG
Reviewed by Adrienne Lin

▲Back to top

ABOUT THE SPEAKER
Michael Rubinstein - Research scientist, Google
Computer scientist Michael Rubinstein and his team have developed a "motion microscope" that can show video footage of barely perceivable movements, like breaths and heartbeats.

Why you should listen

Michael Rubinstein zooms in on what we can't see and mangnifies it by thirty or a hundred times. His "motion microscope," developed at MIT with Microsoft and Quanta Research, picks up on subtle motion and color changes in videos and blows them up for the naked eye to see. The result: fun, cool, creepy videos.

Rubinstein is a research scientist at a new Cambridge-based Google lab for computer vision research. He has a PhD in computer science and electrical engineering from MIT.

More profile about the speaker
Michael Rubinstein | Speaker | TED.com