1
00:00:05,620 --> 00:00:09,939
Welcome back to this another machine learning tutorial. Today we're going to create a project

2
00:00:09,939 --> 00:00:16,800
that is based on sound and in this project we'll train a computer to understand an alien language

3
00:00:16,800 --> 00:00:24,140
or our language and we'll use that we'll use that model of sound and language to control

4
00:00:24,140 --> 00:00:29,660
an alien character. Okay so let's get started. First of all this is your dashboard and once

5
00:00:29,660 --> 00:00:37,899
log in you will see this go to projects and click on add new project okay so this project will be

6
00:00:37,899 --> 00:00:48,409
named as this project will be named as alien language and the recognizing model is sound

7
00:00:50,009 --> 00:00:57,210
okay let's press create after that once you clear once you have created this project you

8
00:00:57,210 --> 00:01:04,010
will just click on this project and go to train okay so this is the background noise block and

9
00:01:04,010 --> 00:01:09,049
we have to give it some examples of our background noise without even speaking so the computer can

10
00:01:09,049 --> 00:01:14,250
recognize which is our sound and which is the background noise okay so we'll just click add

11
00:01:14,250 --> 00:01:27,590
example and stay silent and let the background noise fill the fill these examples make sure that

12
00:01:27,590 --> 00:01:33,530
you add multiple examples I would suggest that you add at least ten okay

13
00:01:33,530 --> 00:01:37,469
so once you are done with ten examples of for background noise we'll create a

14
00:01:37,469 --> 00:01:44,109
new label go to this add new label and we'll create a label for left and in

15
00:01:44,109 --> 00:01:50,950
this label will add examples of us speaking left ten times so the start by

16
00:01:50,950 --> 00:02:05,420
creating ad example click on this button and say left left left left okay so as

17
00:02:05,420 --> 00:02:10,199
you can see that these all examples of left have a pictorical representation

18
00:02:10,199 --> 00:02:14,539
and as you can see that these are quite similar to each other because these are

19
00:02:14,539 --> 00:02:21,120
the representation of sound waves and once you say left then that left will be

20
00:02:21,120 --> 00:02:26,259
similar to the previous left okay so we'll just create a new label and this

21
00:02:26,259 --> 00:02:40,509
time we'll name them right and add the examples for right right right okay so

22
00:02:40,509 --> 00:02:44,229
once we have ten examples each of background noise left and right you can

23
00:02:44,229 --> 00:02:48,990
add as many words and commands as you like but for this example we're gonna go

24
00:02:48,990 --> 00:02:55,169
with left and right okay you can go back to project and go to learn and test and

25
00:02:55,169 --> 00:03:04,259
test this machine learning model train this new machine learning model okay so

26
00:03:04,259 --> 00:03:09,639
it will start training machine learning model and it will take some time so wait

27
00:03:09,639 --> 00:03:15,759
for it and if in a minute it says training from training it says available

28
00:03:15,759 --> 00:03:19,580
then you are ready to go if it takes more than a minute prior refreshing your

29
00:03:19,580 --> 00:03:28,750
page and let's see when this when it's available okay so as you can see that

30
00:03:28,750 --> 00:03:36,110
this model to the training model is now available and it's trained so we can test it by making a

31
00:03:36,110 --> 00:03:43,629
sound and recognizing and seeing that how it is recognized by the computer so we'll click on start

32
00:03:43,629 --> 00:03:49,310
listing and produce a sound and then we'll click on stop listening and then we'll see how the

33
00:03:49,310 --> 00:04:04,319
computer recognizes that okay so you can start listening left okay now we can test this model

34
00:04:04,319 --> 00:04:17,069
by click and start listening left right yeah so it works right it recognizes left as left

35
00:04:17,069 --> 00:04:20,670
and right is right okay so we'll go back to project and we'll create something

36
00:04:21,230 --> 00:04:28,670
using this model and we'll go for that we'll go to make and use scratch 3.