Today we will teach everyone
Automatically generate flash text video with Python
The most famous flash text video
It's Apple's 16 years of advertising
The text flashes with the music rhythm
The effect is simple and shocking
Although it seems to have been badly done in the past two years
I went out for a skewers two days ago and found that the store’s ads were all pop-ups
But if you really want to do it, it’s actually a little troublesome
Whether you use PPT or Premier
You have to slowly align the drum rhythm
I’m afraid to make a video to loop this song hundreds of times
I don’t want to hear this song anymore
The effect is comparable to setting a song as an alarm clock in the morning
Nowadays!
Yuechuang is here to help you hard-working video practitioners
We use omnipotent Python
Automatically generate flash text video!
Even if the boss asks for a two-hour flash text video
Don't be afraid!
First of all, I will try it with a very famous song
The reform breeze blows all over the ground
The original work was created by Xiao Keer at station B
https://www.bilibili.com/video/av19390801
You can click to read the original text to watch
Then we look at the flash text video generated by the program automatically find the drum
In the video above
All drum beats are automatically found by the program
Video generation is also all done by the program
The only manual operation is
Split the text as needed
In fact, this step can also be omitted
For example, the following video
I use the word segmentation module
This news
https://news.163.com/19/0618/20/EHVSFA1J0001875P.html
Use flash text
Show up
The way this news is presented
Isn't it very interesting?
Of course, if every news article reads this way
It's probably exhausting...
If you want to get the Python code of this flash text directly
Can reply in the background: flash text
Next, we briefly explain how this code is implemented step by step
First of all, the modules we need to use are
books
moviepy
jieba
click
among them
librosa is used to analyze drum rhythm
moviepy is used to generate video
jieba is used to segment words
click is used to make command line tools
First import the modules we need
import librosa
from moviepy.editor import TextClip, CompositeVideoClip, AudioFileClip
import jieba
import click
Next, define our main function main()
There are 6 input parameters
width: the width of the generated video
height: the height of the generated video
text: The address of the text file containing the text
music: background music
word_split: Whether to split words automatically (splitting according to line breaks without automatic word splitting)
output: the file name of the generated video
At the same time we use click to help us
Make this python file into a command line tool
The usage of click is as follows
See github for details
https://github.com/pallets/click
So our main() function becomes as follows
@click.command()
@click.option('--width', prompt='Width', default=360, help='The width of video clips')
@click.option('--height', prompt='Height', default=240, help='The height of video clips')
@click.option('--text', prompt='Text file', default='text.txt', help='The source text file')
@click.option('--music', prompt='Music file', default='改革春风吹满地.mp3', help='The music file')
@click.option('--word_split', prompt='Split words', default=False, help='Split words or not')
@click.option('--output', prompt='Output file', default='FlashText.mp4', help='The output file name')
def main(width, height, text, music, word_split, output):
The prompt parameter represents the input prompt that will pop up when running the program
default is the default value
That is, the value selected when the user directly enters
Next in the main() function
We do everything
First read in a text file with text
with open(text, 'r', encoding='utf-8') as f:
text_str = f.read()
Then we see whether to segment words according to the user's choice
if word_split:
seg_list = jieba.lcut(text_str)
punct = set(''':!),.:;?]}¢'"、。〉》」』】〕〗〞︰︱︳﹐､﹒
﹔﹕﹖﹗﹚﹜﹞！），．：；？｜｝︴︶︸︺︼︾﹀﹂﹄﹏､～￠
々‖•·ˇˉ―--′’”([{£¥'"‵〈《「『【〔〖（［｛￡￥〝︵︷︹︻
︽︿﹁﹃﹙﹛﹝（｛“‘-—_…/\\''')
word_list = list(filter(lambda x: x not in punct, seg_list))
else:
word_list = text_str.split('\n')
Here when word_split is True
We use jieba.lcut(text_str) to get the list after word segmentation
Similar to the picture below
Next we use two sentences to eliminate
Punctuation in the list after word segmentation
The most important one is to use the filter() function
Combine lambda to eliminate
It's a relatively simple way to remove symbols
And if word_split is False
Then we directly split the document with line breaks
After finishing dividing the text
Let's analyze the music drums
Here librosa can help a lot
librosa is a very powerful Python module
Various information of audio can be analyzed
Here we use its beat_track method
y, sr = librosa.load(music)
tempo, beats = librosa.beat.beat_track(y=y, sr=sr)
beat_times = list(librosa.frames_to_time(beats, sr=sr))
beat_times.append(beat_times[-1] + 1)
Eventually we will get a list of beat_times
It contains the time of each drum
At the same time you can notice the last line
beat_times.append(beat_times[-1] + 1)
Here I repeat the last digit of the list
And offset 1 second later
Mainly for us to deal with the video clip of the last drum beat later
After getting the drum time list
Let’s start making each text video clip
clips = []
for index, beat_time in enumerate(beat_times[:-1]):
if index >= len(word_list):
break
print(f'{index + 1}/{len(beat_times)}——{word_list[index]}')
text_clip = TextClip(
word_list[index],
fontsize=width // 8,
color='white',
size=(width, height),
method='caption',
font='msyhbd.ttc')\
.set_start(beat_time)\
.set_end(beat_times[index + 1])
text_clip = text_clip.set_pos('center')
clips.append(text_clip)
Here we use moviepy
moviepy is a python module that I personally like very much
A very powerful python video production module
I have the opportunity to talk more about the use of this module in the future
Here we use the moviepy text fragment
TextClip()
The first parameter is the specific text
Then you can set the font size, color, etc.
method='caption' means that the text is a caption
When TextClip is set to the title class
We can set size at the same time
When the text width exceeds size, it will wrap automatically
Then you can directly call the two methods in a chain
set_start()和set_end()
Used to set the start and end time of this video clip
Here we just set it as the time between a certain drum beat and the next drum beat
So we have completed the production of each text fragment video
Finally, to generate a complete video
final_clip = CompositeVideoClip(clips)
audio_clip = AudioFileClip(music)
final_video = final_clip.set_audio(audio_clip)
final_video.write_videofile(
output,
fps=30,
codec='mpeg4',
preset='ultrafast',
audio_codec="libmp3lame",
threads=4)
Use moviepy's CompositeVideoClip
To synthesize multiple video clips
Use AudioFileClip to make audio clips
Then synthesize
Finally, directly use the write_videofile() method to generate a complete video
Video and audio encoding can be set in this method
And frame rate FPS
Finally add
if __name__ == '__main__':
main()
That's it!
So I want to make a simple flash text video in the future
Just run this python file
Enter information as prompted
You can wait for the program to do it for you
Is it super convenient?
This issue of Yuechuang Small Classroom is over here
Welcome to follow AI Yuechuang
I will continue to teach you some fun and interesting brain projects