Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -9,7 +9,109 @@ import os
|
|
| 9 |
import requests
|
| 10 |
import tempfile
|
| 11 |
|
| 12 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
|
| 14 |
def main():
|
| 15 |
st.set_page_config(page_title="AI Voiceover", page_icon="🔮")
|
|
|
|
| 9 |
import requests
|
| 10 |
import tempfile
|
| 11 |
|
| 12 |
+
# Load environment variables from .env.local
|
| 13 |
+
load_dotenv('.env.local')
|
| 14 |
+
|
| 15 |
+
def check_password():
|
| 16 |
+
correct_password = os.getenv('PASSWORD')
|
| 17 |
+
if correct_password is None:
|
| 18 |
+
st.error("Password is not set in .env.local")
|
| 19 |
+
return False
|
| 20 |
+
|
| 21 |
+
user_password = st.text_input("Enter the password to proceed", type="password")
|
| 22 |
+
if user_password == correct_password:
|
| 23 |
+
return True
|
| 24 |
+
else:
|
| 25 |
+
if st.button("Check Password"):
|
| 26 |
+
st.error("Incorrect password")
|
| 27 |
+
return False
|
| 28 |
+
|
| 29 |
+
def video_to_frames(video_file, frame_sampling_rate=1):
|
| 30 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix='.mp4') as tmpfile:
|
| 31 |
+
tmpfile.write(video_file.read())
|
| 32 |
+
video_filename = tmpfile.name
|
| 33 |
+
|
| 34 |
+
video_clip = VideoFileClip(video_filename)
|
| 35 |
+
video_duration = video_clip.duration
|
| 36 |
+
fps = video_clip.fps
|
| 37 |
+
frames_to_skip = int(fps * frame_sampling_rate)
|
| 38 |
+
|
| 39 |
+
video = cv2.VideoCapture(video_filename)
|
| 40 |
+
base64Frame = []
|
| 41 |
+
current_frame = 0
|
| 42 |
+
|
| 43 |
+
while video.isOpened():
|
| 44 |
+
success, frame = video.read()
|
| 45 |
+
if not success:
|
| 46 |
+
break
|
| 47 |
+
if current_frame % frames_to_skip == 0:
|
| 48 |
+
_, buffer = cv2.imencode('.jpg', frame)
|
| 49 |
+
base64Frame.append(base64.b64encode(buffer).decode("utf-8"))
|
| 50 |
+
current_frame += 1
|
| 51 |
+
|
| 52 |
+
video.release()
|
| 53 |
+
print(f"{len(base64Frame)} frames read at a sampling rate of {frame_sampling_rate} second(s) per frame.")
|
| 54 |
+
return base64Frame, video_filename, video_duration
|
| 55 |
+
|
| 56 |
+
def frames_to_story(base64Frames, prompt, api_key):
|
| 57 |
+
PROMPT_MESSAGES = [
|
| 58 |
+
{
|
| 59 |
+
"role": "user",
|
| 60 |
+
"content": [
|
| 61 |
+
prompt,
|
| 62 |
+
*map(lambda x: {"image": x, "resize": 768}, base64Frames[0::50]),
|
| 63 |
+
],
|
| 64 |
+
},
|
| 65 |
+
]
|
| 66 |
+
params = {
|
| 67 |
+
"model": "gpt-4-vision-preview",
|
| 68 |
+
"messages": PROMPT_MESSAGES,
|
| 69 |
+
"api_key": api_key,
|
| 70 |
+
"headers": {"Openai-Version": "2020-11-07"},
|
| 71 |
+
"max_tokens": 1000,
|
| 72 |
+
}
|
| 73 |
+
result = openai.ChatCompletion.create(**params)
|
| 74 |
+
print(result.choices[0].message.content)
|
| 75 |
+
return result.choices[0].message.content
|
| 76 |
+
|
| 77 |
+
def text_to_audio(text, api_key, voice):
|
| 78 |
+
response = requests.post(
|
| 79 |
+
"https://api.openai.com/v1/audio/speech",
|
| 80 |
+
headers={
|
| 81 |
+
"Authorization": f"Bearer {api_key}",
|
| 82 |
+
},
|
| 83 |
+
json={
|
| 84 |
+
"model": "tts-1",
|
| 85 |
+
"input": text,
|
| 86 |
+
"voice": voice,
|
| 87 |
+
},
|
| 88 |
+
)
|
| 89 |
+
|
| 90 |
+
if response.status_code != 200:
|
| 91 |
+
raise Exception("Request failed with status code")
|
| 92 |
+
|
| 93 |
+
audio_bytes_io = io.BytesIO()
|
| 94 |
+
for chunk in response.iter_content(chunk_size=1024*1024):
|
| 95 |
+
audio_bytes_io.write(chunk)
|
| 96 |
+
audio_bytes_io.seek(0)
|
| 97 |
+
|
| 98 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as tmpfile:
|
| 99 |
+
for chunk in response.iter_content(chunk_size=1024*1024):
|
| 100 |
+
tmpfile.write(chunk)
|
| 101 |
+
audio_filename = tmpfile.name
|
| 102 |
+
|
| 103 |
+
return audio_filename, audio_bytes_io
|
| 104 |
+
|
| 105 |
+
def merge_audio_video(video_filename, audio_filename, output_filename):
|
| 106 |
+
print("Merging audio and video ...")
|
| 107 |
+
video_clip = VideoFileClip(video_filename)
|
| 108 |
+
audio_clip = AudioFileClip(audio_filename)
|
| 109 |
+
final_clip = video_clip.set_audio(audio_clip)
|
| 110 |
+
final_clip.write_videofile(output_filename, codec='libx264', audio_codec="aac")
|
| 111 |
+
video_clip.close()
|
| 112 |
+
audio_clip.close()
|
| 113 |
+
return output_filename
|
| 114 |
+
|
| 115 |
|
| 116 |
def main():
|
| 117 |
st.set_page_config(page_title="AI Voiceover", page_icon="🔮")
|