Paranoid Android
Elite Member
- Jun 20, 2010
- 2,864
- 4,209
Do you look fugly even with filters? Do you sound like a croaking toad? Are you camera shy to make videos of yourself? Or are you just so damn lazy to use a video editor to make your own shorts from stock footage? Are you having to pay people on Fiverr to do it for you? Paranoid Android on Beast Mode is here for your rescue.
Get your API key from OpenAI.com, sign up and get a free $18 credit with the key. This is plenty to get you started
Install pycharm or your favorite python IDE and paste the script below.
Install all the libraries, pycharm makes it easy if you don't know what you're doing.
Run the script.
Order of execution:
1. Download background videos first. You can pick from copyright free playlists, single videos, or from a built in list of 4 videos.
2. Create Audio. Pass in a well formatted question about the topic, or make a list in a file q.txt and let OpenAI generate the text for you, which is then converted to voice files with Google Text to Speech
3. Make videos from the created audio files.
4. Upload to youtube or tiktok and watch the money roll in
If the Audio length is less than 60 secs, video is made to that length, and if the audio is longer than 60 secs, it gets trimmed to 60 secs. Play around with openai for a bit to figure out the optimal question format for what you're asking if you don't want too much of valuable information cut off in the end.
Audio creation is multithreaded so it is all done in almost no time if you have an ssd, and the background video downloads (required once) will depend on your internet speeds. But the videomaking at 1080x1920p resolution takes about 15 minutes per video on my 8 core Ryzen laptop with 16 gigs of ram running linux. So if you're going to be making multiple videos, you're going to have to leave your computer running when you're sleeping.
Wanna make your videos with your own audio? I've seen people do it with funny call recordings and stuff on TikTok. Just put your audio inside the 'tts' folder (after running the script once so the folders are created), and hit option 3 to get the audio cropped to 60 seconds and videos of exactly 1800 frames.
Feature requests? Bugs reports? Appreciation? Please post them here.
Like we say in the south of India, Yenjoy!
P.S. The bot can be a little rude at times.
Get your API key from OpenAI.com, sign up and get a free $18 credit with the key. This is plenty to get you started
Install pycharm or your favorite python IDE and paste the script below.
Install all the libraries, pycharm makes it easy if you don't know what you're doing.
Run the script.
Order of execution:
1. Download background videos first. You can pick from copyright free playlists, single videos, or from a built in list of 4 videos.
2. Create Audio. Pass in a well formatted question about the topic, or make a list in a file q.txt and let OpenAI generate the text for you, which is then converted to voice files with Google Text to Speech
3. Make videos from the created audio files.
4. Upload to youtube or tiktok and watch the money roll in
If the Audio length is less than 60 secs, video is made to that length, and if the audio is longer than 60 secs, it gets trimmed to 60 secs. Play around with openai for a bit to figure out the optimal question format for what you're asking if you don't want too much of valuable information cut off in the end.
Audio creation is multithreaded so it is all done in almost no time if you have an ssd, and the background video downloads (required once) will depend on your internet speeds. But the videomaking at 1080x1920p resolution takes about 15 minutes per video on my 8 core Ryzen laptop with 16 gigs of ram running linux. So if you're going to be making multiple videos, you're going to have to leave your computer running when you're sleeping.
Wanna make your videos with your own audio? I've seen people do it with funny call recordings and stuff on TikTok. Just put your audio inside the 'tts' folder (after running the script once so the folders are created), and hit option 3 to get the audio cropped to 60 seconds and videos of exactly 1800 frames.
Python:
import concurrent.futures
import openai, os, random, time
from gtts import gTTS
import moviepy.editor as mp
from mutagen.mp3 import MP3
from pytube import Playlist, YouTube
from moviepy.video.io.VideoFileClip import VideoFileClip
from moviepy.audio.io.AudioFileClip import AudioFileClip
openai.api_key = "Your Key Here"
videoHeight = 1920
audiopath = 'tts/'
videopath = 'finalvideo/'
def audioMaker(i, q):
q = q.strip('\n')
response = openai.Completion.create(
model="text-davinci-002",
prompt=q,
temperature=0.7,
max_tokens=500,
top_p=1.0,
frequency_penalty=0.0,
presence_penalty=0.0
)
print(q, 'done')
tts = gTTS(text=response["choices"][0]["text"], tld='ca', slow=False)
if os.path.exists(f"tts/{q.replace(' ', '_')}.mp3"):
newfilename = '_' + q.replace(' ', '_') + '.mp3'
tts.save(f"tmp/{newfilename.replace(' ', '_')}.mp3")
os.replace(f"tmp/{newfilename.replace(' ', '_')}.mp3", f"tts/{newfilename.replace(' ', '_')}.mp3")
print('iteration', i, q, 'saved')
else:
tts.save(f"tmp/{q.replace(' ', '_')}.mp3")
os.replace(f"tmp/{q.replace(' ', '_')}.mp3", f"tts/{q.replace(' ', '_')}.mp3")
print('iteration', i, q, 'saved')
def videoMaker():
aflist = os.listdir(path='tts/')
for i, fname in enumerate(aflist):
retry = 0
if fname.endswith('.mp3'):
while True:
randomVideo = random.choice(os.listdir('bgvideo/'))
if randomVideo.endswith('.mp4'):
backgroundVideo = VideoFileClip('bgvideo/' + randomVideo).without_audio().resize(height=videoHeight).crop(x1=1166.6, y1=0, x2=2246.6, y2=1920)
audioFileDuration = MP3(audiopath + fname).info.length
if audioFileDuration < 60:
audioDuration = audioFileDuration
audioClip = mp.CompositeAudioClip([AudioFileClip(audiopath + fname)])
print('video length is shorter than 60 secs', audioDuration)
else:
audioDuration = 60
audioClip = mp.CompositeAudioClip([AudioFileClip(audiopath + fname)]).subclip(0, 60)
videoDuration = backgroundVideo.duration
if videoDuration >= audioDuration:
randRange = random.randrange(30, int(videoDuration - audioDuration))
videoSubClip = backgroundVideo.subclip(randRange, randRange + audioDuration)
print('found a video longer than audio')
videoSubClip.audio = audioClip
videoSubClip.write_videofile('tmp/' + fname.replace('.mp3', '.mp4'), fps=30, audio_codec='aac',
audio_bitrate='192k', verbose=True, threads=os.cpu_count())
os.replace(audiopath + fname, 'usedaudio/' + fname)
os.replace('tmp/'+ fname.replace('.mp3', '.mp4'), videopath + fname.replace('.mp3', '.mp4'))
break
else:
retry += 1
print('video is shorter than your fucking dick, retrying : ', retry + 1)
if retry > 10:
print('Audio probably too long, skipping after 10 tries. Skipping audio', fname)
os.replace(audiopath + fname, 'audiotoolong/' + fname)
break
def backgroundVideoDownloader():
opt = int(input('1. Playlist\n2.Video\n3. Download built in List of Videos Or any other key to return\n Wachawannado? : '))
if opt == 1:
p = Playlist(input('What is the playlist link? : '))
n = int(input('How many videos do you want to download?\n Hit 0 for all : '))
for i, video in enumerate(p.videos):
vidtitle = str(random.randint(100000, 1000000)) + '.mp4'
print('Downloading', vidtitle)
video.streams.filter(res="1080p").first().download(output_path='tmp/', filename=vidtitle, timeout=30,
max_retries=3, skip_existing=False)
if os.path.exists('bgvideo/' + vidtitle):
altvidtitle = str(random.randint(100000, 1000000)) + '.mp4'
os.replace('tmp/' + vidtitle, 'bgvideo/' + altvidtitle)
else:
os.replace('tmp/' + vidtitle, 'bgvideo/' + vidtitle)
print(i + 1, video.title, 'downloaded as', vidtitle)
if i + 1 == n:
break
elif opt == 2:
vidtitle = str(random.randint(100000, 1000000)) + '.mp4'
p = YouTube(input('What is the video link? : '))
p.streams.filter(res="1080p").first().download(output_path='tmp/', filename=vidtitle, skip_existing=False)
if os.path.exists('bgvideo/' + vidtitle):
altvidtitle = str(random.randint(100000, 1000000)) + '.mp4'
os.replace('tmp/' + vidtitle, 'bgvideo/' + altvidtitle)
print(p.title, 'downloaded as', altvidtitle)
else:
os.replace('tmp/' + vidtitle, 'bgvideo/' + vidtitle)
print(p.title, 'downloaded as', vidtitle)
print(p.title, 'downloaded')
elif opt == 3:
builtInListOfVideos = ["https://www.youtube.com/watch?v=vw5L4xCPy9Q", "https://www.youtube.com/watch?v=2X9QGY__0II", "https://www.youtube.com/watch?v=n_Dv4JMiwK8", "https://www.youtube.com/watch?v=qGa9kWREOnE"]
for v in builtInListOfVideos:
vidtitle = str(random.randint(100000, 1000000)) + '.mp4'
YouTube(v).streams.filter(res="1080p").first().download(output_path='tmp/', filename=vidtitle, skip_existing=False)
if os.path.exists('bgvideo/' + vidtitle):
altvidtitle = str(random.randint(100000, 1000000)) + '.mp4'
os.replace('tmp/' + vidtitle, 'bgvideo/' + altvidtitle)
print(p.title, 'downloaded as', altvidtitle)
else:
os.replace('tmp/' + vidtitle, 'bgvideo/' + vidtitle)
print(p.title, 'downloaded as', vidtitle)
print(YouTube(v).title, 'downloaded')
else:
print('returning to main menu')
return
if __name__ == '__main__':
if not os.path.exists(audiopath):
os.mkdir('tts')
if not os.path.exists('audiotoolong/'):
os.mkdir('audiotoolong')
if not os.path.exists('bgvideo/'):
os.mkdir('bgvideo')
if not os.path.exists(videopath):
os.mkdir('finalvideo')
if not os.path.exists('tmp/'):
os.mkdir('tmp')
if not os.path.exists('usedaudio/'):
os.mkdir('usedaudio')
try:
while True:
wachawannado = int(input(
'1. Make Audio from list in conditions.txt\n2. Input conditions manually to make audio\n3. Make videos from available audio\n4. Download new backgrounds\n5. Be the quitter that you are\nWachawannado? : '))
if wachawannado == 1:
if os.path.exists('q.txt'):
with open('q.txt', 'r') as q:
with concurrent.futures.ProcessPoolExecutor(max_workers=64) as executor:
for i, lines in enumerate(q):
executor.submit(audioMaker, i=i, q=lines)
else:
print('wheres the fucking file? Create q.txt or hit 2 in the main menu')
elif wachawannado == 2:
audioMaker(q=input('Type a well worded clearly formatted question: '), i=1)
elif wachawannado == 3:
videoMaker()
elif wachawannado == 4:
backgroundVideoDownloader()
elif wachawannado == 5:
print('fuckin quitters')
break
else:
print('\nWTF, try that again\n')
except KeyboardInterrupt:
tmpdir = os.listdir('tmp/')
if not tmpdir:
print('\ntmp folder is empty, now fuck off')
else:
for i, file in enumerate(tmpdir):
os.remove('tmp/' + file)
print(f'\n{i + 1} files cleared from the tmp folder. now fuck off')Feature requests? Bugs reports? Appreciation? Please post them here.
Like we say in the south of India, Yenjoy!
P.S. The bot can be a little rude at times.
