1
00:00:00,180 --> 00:00:02,610
So now let's talk about Amazon Transcribe.

2
00:00:02,610 --> 00:00:04,170
So Transcribe, as the name indicates,

3
00:00:04,170 --> 00:00:07,290
is used to convert speech into text.

4
00:00:07,290 --> 00:00:09,180
So I will speak into my microphone

5
00:00:09,180 --> 00:00:12,750
and this will be the audiogram I get out of it.

6
00:00:12,750 --> 00:00:15,450
And then, thanks to Transcribe, it's going to analyze

7
00:00:15,450 --> 00:00:17,827
the text and say, "Hello, my name is Stephane.

8
00:00:17,827 --> 00:00:19,710
"I hope you're enjoying the course!"

9
00:00:19,710 --> 00:00:21,240
Okay, so how does it do it?

10
00:00:21,240 --> 00:00:24,000
Well, it uses a deep learning process called

11
00:00:24,000 --> 00:00:27,930
automatic speak recognition, ASR, to convert the speech

12
00:00:27,930 --> 00:00:29,883
to text quickly and accurately.

13
00:00:30,720 --> 00:00:35,430
And what if there is some PII data in the audio?

14
00:00:35,430 --> 00:00:39,000
So PII means personally identifiable information.

15
00:00:39,000 --> 00:00:41,250
For example, someone's age and so on.

16
00:00:41,250 --> 00:00:43,350
Then we can use the redaction feature

17
00:00:43,350 --> 00:00:45,360
from within Amazon Transcribe.

18
00:00:45,360 --> 00:00:48,450
And so thanks to it, we can remove automatically

19
00:00:48,450 --> 00:00:50,580
any PII data using redaction.

20
00:00:50,580 --> 00:00:52,747
So it will say, "Hello, my name is Stephane.

21
00:00:52,747 --> 00:00:54,960
"My age is," and then blank

22
00:00:54,960 --> 00:00:57,270
because we've removed that information.

23
00:00:57,270 --> 00:00:59,220
So the use cases for Amazon Transcribe

24
00:00:59,220 --> 00:01:01,470
is to transcribe customer service calls

25
00:01:01,470 --> 00:01:04,379
or to automate closed captioning and subtitling

26
00:01:04,379 --> 00:01:07,140
and to generate metadata for media assets

27
00:01:07,140 --> 00:01:10,173
in order to create a fully searchable archive.

28
00:01:11,100 --> 00:01:13,560
If you wanted, you could try out Transcribe directly

29
00:01:13,560 --> 00:01:15,390
from the AWS console.

30
00:01:15,390 --> 00:01:18,030
You can launch Transcribe and then you would,

31
00:01:18,030 --> 00:01:19,500
for example, start streaming.

32
00:01:19,500 --> 00:01:24,420
So we'd say, "Hello, I really like this course,"

33
00:01:24,420 --> 00:01:26,430
then stop streaming and then you would get

34
00:01:26,430 --> 00:01:27,570
the transcription outputs.

35
00:01:27,570 --> 00:01:30,090
So this is a good way to see how Transcribe would work.

36
00:01:30,090 --> 00:01:32,490
Have fun and I will see you in the next lecture.