1
00:00:00,000 --> 00:00:01,650
In the previous video,

2
00:00:01,650 --> 00:00:04,080
we used the single song
to generate text.

3
00:00:04,080 --> 00:00:06,450
We got some text
from it but once we

4
00:00:06,450 --> 00:00:08,640
tried to predict
beyond a few words,

5
00:00:08,640 --> 00:00:11,025
it rapidly became gibberish.

6
00:00:11,025 --> 00:00:12,570
So in this video,

7
00:00:12,570 --> 00:00:15,150
we'll take a look at
adapting that work for

8
00:00:15,150 --> 00:00:18,675
a larger body of words
to see the impact.

9
00:00:18,675 --> 00:00:20,400
The good news is that it will

10
00:00:20,400 --> 00:00:22,410
require very little code changes,

11
00:00:22,410 --> 00:00:25,630
so you'll be able to get
it working quite quickly.

12
00:00:26,360 --> 00:00:31,950
I've prepared a file with
a lot of songs that has 1,692

13
00:00:31,950 --> 00:00:34,770
sentences in all to see
what the impact would

14
00:00:34,770 --> 00:00:37,890
be on the poetry that
a neural network would create.

15
00:00:37,890 --> 00:00:41,110
To download these lyrics,
you can use this code.