mohsin / app.py
AZILS's picture
Upload 3 files
c2552dd verified
import os
import re
import json
import time
import random
import tempfile
import requests
import numpy as np
import uuid
from PIL import Image, ImageDraw, ImageFont
from io import BytesIO
from datetime import datetime
import gradio as gr
from dotenv import load_dotenv
import moviepy.editor as mpy
from moviepy.editor import *
from moviepy.audio.fx.all import volumex
from moviepy.video.fx.all import crop
# Suppress the asyncio "Event loop is closed" warning on Windows
import sys
if sys.platform.startswith('win'):
import asyncio
asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())
# Load environment variables from .env file if present
load_dotenv()
# Directory structure constants
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
STATIC_DIR = os.path.join(BASE_DIR, "static")
MUSIC_DIR = os.path.join(STATIC_DIR, "music")
FONTS_DIR = os.path.join(STATIC_DIR, "fonts")
STORAGE_DIR = os.path.join(BASE_DIR, "storage")
# Create necessary directories
os.makedirs(STATIC_DIR, exist_ok=True)
os.makedirs(MUSIC_DIR, exist_ok=True)
os.makedirs(FONTS_DIR, exist_ok=True)
os.makedirs(STORAGE_DIR, exist_ok=True)
# Helper functions for logging
def info(message):
timestamp = datetime.now().strftime("%H:%M:%S")
formatted_message = f"[{timestamp}] [INFO] {message}"
print(formatted_message)
return formatted_message
def success(message):
timestamp = datetime.now().strftime("%H:%M:%S")
formatted_message = f"[{timestamp}] [SUCCESS] {message}"
print(formatted_message)
return formatted_message
def warning(message):
timestamp = datetime.now().strftime("%H:%M:%S")
formatted_message = f"[{timestamp}] [WARNING] {message}"
print(formatted_message)
return formatted_message
def error(message):
timestamp = datetime.now().strftime("%H:%M:%S")
formatted_message = f"[{timestamp}] [ERROR] {message}"
print(formatted_message)
return formatted_message
def get_music_files():
"""Get list of available music files in the music directory."""
if not os.path.exists(MUSIC_DIR):
return ["none"]
music_files = [f for f in os.listdir(MUSIC_DIR) if f.endswith(('.mp3', '.wav'))]
if not music_files:
return ["none"]
return ["random"] + music_files
def get_font_files():
"""Get list of available font files in the fonts directory."""
if not os.path.exists(FONTS_DIR):
return ["default"]
font_files = [f.split('.')[0] for f in os.listdir(FONTS_DIR) if f.endswith(('.ttf', '.otf'))]
if not font_files:
return ["default"]
return ["random"] + font_files
def choose_random_music():
"""Selects a random music file from the music directory."""
if not os.path.exists(MUSIC_DIR):
error(f"Music directory {MUSIC_DIR} does not exist")
return None
music_files = [f for f in os.listdir(MUSIC_DIR) if f.endswith(('.mp3', '.wav'))]
if not music_files:
warning(f"No music files found in {MUSIC_DIR}")
return None
return os.path.join(MUSIC_DIR, random.choice(music_files))
def choose_random_font():
"""Selects a random font file from the fonts directory."""
if not os.path.exists(FONTS_DIR):
error(f"Fonts directory {FONTS_DIR} does not exist")
return "default"
font_files = [f for f in os.listdir(FONTS_DIR) if f.endswith(('.ttf', '.otf'))]
if not font_files:
warning(f"No font files found in {FONTS_DIR}")
return None
return font_files[0].split('.')[0] if len(font_files) == 1 else random.choice([f.split('.')[0] for f in font_files])
class YouTube:
def __init__(self, niche: str, language: str,
text_gen="g4f", text_model="gpt-4",
image_gen="g4f", image_model="flux",
tts_engine="edge", tts_voice="en-US-AriaNeural",
subtitle_font="default", font_size=80,
text_color="white", highlight_color="blue",
subtitles_enabled=True, highlighting_enabled=True,
subtitle_position="bottom", music_file="random",
enable_music=True, music_volume=0.1,
api_keys=None, progress=gr.Progress()) -> None:
"""Initialize the YouTube Shorts Generator."""
self.progress = progress
self.progress(0, desc="Initializing")
# Store basic parameters
info(f"Initializing YouTube class")
self._niche = niche
self._language = language
self.text_gen = text_gen
self.text_model = text_model
self.image_gen = image_gen
self.image_model = image_model
self.tts_engine = tts_engine
self.tts_voice = tts_voice
self.subtitle_font = subtitle_font
self.font_size = font_size
self.text_color = text_color
self.highlight_color = highlight_color
self.subtitles_enabled = subtitles_enabled
self.highlighting_enabled = highlighting_enabled
self.subtitle_position = subtitle_position
self.music_file = music_file
self.enable_music = enable_music
self.music_volume = music_volume
self.api_keys = api_keys or {}
self.images = []
self.logs = []
# Set API keys from parameters or environment variables
if 'gemini' in self.api_keys and self.api_keys['gemini']:
os.environ["GEMINI_API_KEY"] = self.api_keys['gemini']
if 'assemblyai' in self.api_keys and self.api_keys['assemblyai']:
os.environ["ASSEMBLYAI_API_KEY"] = self.api_keys['assemblyai']
if 'elevenlabs' in self.api_keys and self.api_keys['elevenlabs']:
os.environ["ELEVENLABS_API_KEY"] = self.api_keys['elevenlabs']
if 'segmind' in self.api_keys and self.api_keys['segmind']:
os.environ["SEGMIND_API_KEY"] = self.api_keys['segmind']
if 'openai' in self.api_keys and self.api_keys['openai']:
os.environ["OPENAI_API_KEY"] = self.api_keys['openai']
info(f"Niche: {niche}, Language: {language}")
self.log(f"Initialized with niche: {niche}, language: {language}")
self.log(f"Text generator: {text_gen} - Model: {text_model}")
self.log(f"Image generator: {image_gen} - Model: {image_model}")
self.log(f"TTS engine: {tts_engine} - Voice: {tts_voice}")
self.log(f"Subtitles: {'Enabled' if subtitles_enabled else 'Disabled'} - Highlighting: {'Enabled' if highlighting_enabled else 'Disabled'}")
self.log(f"Music: {music_file}")
def log(self, message):
"""Add a log message to the logs list."""
timestamp = datetime.now().strftime("%H:%M:%S")
log_entry = f"[{timestamp}] {message}"
self.logs.append(log_entry)
return log_entry
@property
def niche(self) -> str:
return self._niche
@property
def language(self) -> str:
return self._language
def generate_response(self, prompt: str, model: str = None) -> str:
"""Generate a response using the selected text generation model."""
self.log(f"Generating response for prompt: {prompt[:50]}...")
try:
if self.text_gen == "gemini":
self.log("Using Google's Gemini model")
# Check if API key is set
gemini_api_key = os.environ.get("GEMINI_API_KEY", "")
if not gemini_api_key:
raise ValueError("Gemini API key is not set. Please provide a valid API key.")
import google.generativeai as genai
genai.configure(api_key=gemini_api_key)
model_to_use = model if model else self.text_model
genai_model = genai.GenerativeModel(model_to_use)
response = genai_model.generate_content(prompt).text
elif self.text_gen == "g4f":
self.log("Using G4F for text generation")
import g4f
model_to_use = model if model else self.text_model
self.log(f"Using G4F model: {model_to_use}")
response = g4f.ChatCompletion.create(
model=model_to_use,
messages=[{"role": "user", "content": prompt}]
)
elif self.text_gen == "openai":
self.log("Using OpenAI for text generation")
openai_api_key = os.environ.get("OPENAI_API_KEY", "")
if not openai_api_key:
raise ValueError("OpenAI API key is not set. Please provide a valid API key.")
from openai import OpenAI
client = OpenAI(api_key=openai_api_key)
model_to_use = model if model else "gpt-3.5-turbo"
response = client.chat.completions.create(
model=model_to_use,
messages=[{"role": "user", "content": prompt}]
).choices[0].message.content
else:
# No fallback, raise an exception for unsupported text generator
error_msg = f"Unsupported text generator: {self.text_gen}"
self.log(error(error_msg))
raise ValueError(error_msg)
self.log(f"Response generated successfully, length: {len(response)} characters")
return response
except Exception as e:
error_msg = f"Error generating response: {str(e)}"
self.log(error(error_msg))
raise Exception(error_msg)
def generate_topic(self) -> str:
"""Generate a topic based on the YouTube Channel niche."""
self.progress(0.05, desc="Generating topic")
self.log("Generating topic based on niche")
completion = self.generate_response(
f"Please generate a specific video idea that takes about the following topic: {self.niche}. "
f"Make it exactly one sentence. Only return the topic, nothing else."
)
if not completion:
self.log(error("Failed to generate Topic."))
raise Exception("Failed to generate a topic. Please try again with a different niche.")
self.subject = completion
self.log(success(f"Generated topic: {completion}"))
return completion
def generate_script(self) -> str:
"""Generate a script for a video, based on the subject and language."""
self.progress(0.1, desc="Creating script")
self.log("Generating script for video")
prompt = f"""
Generate a script for youtube shorts video, depending on the subject of the video.
The script is to be returned as a string with the specified number of paragraphs.
Here is an example of a string:
"This is an example string."
Do not under any circumstance reference this prompt in your response.
Get straight to the point, don't start with unnecessary things like, "welcome to this video".
Obviously, the script should be related to the subject of the video.
YOU MUST NOT INCLUDE ANY TYPE OF MARKDOWN OR FORMATTING IN THE SCRIPT, NEVER USE A TITLE.
YOU MUST WRITE THE SCRIPT IN THE LANGUAGE SPECIFIED IN [LANGUAGE].
ONLY RETURN THE RAW CONTENT OF THE SCRIPT. DO NOT INCLUDE "VOICEOVER", "NARRATOR" OR SIMILAR INDICATORS.
Subject: {self.subject}
Language: {self.language}
"""
completion = self.generate_response(prompt)
# Apply regex to remove *
completion = re.sub(r"\*", "", completion)
if not completion:
self.log(error("The generated script is empty."))
raise Exception("Failed to generate a script. Please try again.")
if len(completion) > 5000:
self.log(warning("Generated script is too long."))
raise ValueError("Generated script exceeds 5000 characters. Please try again.")
self.script = completion
self.log(success(f"Generated script ({len(completion)} chars)"))
return completion
def generate_metadata(self) -> dict:
"""Generate video metadata (title, description)."""
self.progress(0.15, desc="Creating title and description")
self.log("Generating metadata (title and description)")
title = self.generate_response(
f"Please generate a YouTube Video Title for the following subject, including hashtags: "
f"{self.subject}. Only return the title, nothing else. Limit the title under 100 characters."
)
if len(title) > 100:
self.log(warning("Generated title exceeds 100 characters."))
raise ValueError("Generated title exceeds 100 characters. Please try again.")
description = self.generate_response(
f"Please generate a YouTube Video Description for the following script: {self.script}. "
f"Only return the description, nothing else."
)
self.metadata = {
"title": title,
"description": description
}
self.log(success(f"Generated title: {title}"))
self.log(success(f"Generated description: {description[:50]}..."))
return self.metadata
def generate_prompts(self, count=5) -> list:
"""Generate AI Image Prompts based on the provided Video Script."""
self.progress(0.2, desc="Creating image prompts")
self.log(f"Generating {count} image prompts")
prompt = f"""
Generate {count} Image Prompts for AI Image Generation,
depending on the subject of a video.
Subject: {self.subject}
The image prompts are to be returned as
a JSON-Array of strings.
Each search term should consist of a full sentence,
always add the main subject of the video.
Be emotional and use interesting adjectives to make the
Image Prompt as detailed as possible.
YOU MUST ONLY RETURN THE JSON-ARRAY OF STRINGS.
YOU MUST NOT RETURN ANYTHING ELSE.
YOU MUST NOT RETURN THE SCRIPT.
The search terms must be related to the subject of the video.
Here is an example of a JSON-Array of strings:
["image prompt 1", "image prompt 2", "image prompt 3"]
For context, here is the full text:
{self.script}
"""
completion = str(self.generate_response(prompt))\
.replace("```json", "") \
.replace("```", "")
image_prompts = []
if "image_prompts" in completion:
try:
image_prompts = json.loads(completion)["image_prompts"]
except:
self.log(warning("Failed to parse 'image_prompts' from JSON response."))
if not image_prompts:
try:
image_prompts = json.loads(completion)
self.log(f"Parsed image prompts from JSON response.")
except Exception:
self.log(warning("JSON parsing failed. Attempting to extract array using regex..."))
# Get everything between [ and ], and turn it into a list
r = re.compile(r"\[.*\]", re.DOTALL)
matches = r.findall(completion)
if len(matches) == 0:
self.log(warning("Failed to extract array. Unable to create image prompts."))
raise ValueError("Failed to generate valid image prompts. Please try again.")
else:
try:
image_prompts = json.loads(matches[0])
except:
self.log(error("Failed to parse array from regex match."))
# Use regex to extract individual strings
string_pattern = r'"([^"]*)"'
strings = re.findall(string_pattern, matches[0])
if strings:
image_prompts = strings
else:
self.log(error("Failed to extract strings from regex match."))
raise ValueError("Failed to parse image prompts. Please try again.")
# Ensure we have the requested number of prompts
if len(image_prompts) < count:
self.log(warning(f"Received fewer prompts ({len(image_prompts)}) than requested ({count})."))
raise ValueError(f"Received only {len(image_prompts)} prompts instead of {count}. Please try again.")
# Limit to the requested count
image_prompts = image_prompts[:count]
self.image_prompts = image_prompts
self.log(success(f"Generated {len(self.image_prompts)} Image Prompts"))
for i, prompt in enumerate(self.image_prompts):
self.log(f"Image Prompt {i+1}: {prompt}")
return image_prompts
def generate_image(self, prompt) -> str:
"""Generate an image using the selected image generation model."""
self.log(f"Generating image for prompt: {prompt[:50]}...")
# Always save images directly to the generation folder when it exists
if hasattr(self, 'generation_folder') and os.path.exists(self.generation_folder):
image_path = os.path.join(self.generation_folder, f"img_{uuid.uuid4()}_{int(time.time())}.png")
else:
# Use STORAGE_DIR if no generation folder
image_path = os.path.join(STORAGE_DIR, f"img_{uuid.uuid4()}_{int(time.time())}.png")
if self.image_gen == "prodia":
self.log("Using Prodia provider for image generation")
s = requests.Session()
headers = {
"user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"
}
# Generate job
self.log("Sending generation request to Prodia API")
resp = s.get(
"https://api.prodia.com/generate",
params={
"new": "true",
"prompt": prompt,
"model": self.image_model,
"negative_prompt": "verybadimagenegative_v1.3",
"steps": "20",
"cfg": "7",
"seed": random.randint(1, 10000),
"sample": "DPM++ 2M Karras",
"aspect_ratio": "square"
},
headers=headers
)
if resp.status_code != 200:
raise Exception(f"Prodia API error: {resp.text}")
job_id = resp.json()['job']
self.log(f"Job created with ID: {job_id}")
# Wait for generation to complete
max_attempts = 30
attempts = 0
while attempts < max_attempts:
attempts += 1
time.sleep(2)
status = s.get(f"https://api.prodia.com/job/{job_id}", headers=headers).json()
if status["status"] == "succeeded":
self.log("Image generation successful, downloading result")
img_data = s.get(f"https://images.prodia.xyz/{job_id}.png?download=1", headers=headers).content
with open(image_path, "wb") as f:
f.write(img_data)
self.images.append(image_path)
self.log(success(f"Image saved to: {image_path}"))
return image_path
elif status["status"] == "failed":
raise Exception(f"Prodia job failed: {status.get('error', 'Unknown error')}")
# Still processing
self.log(f"Still processing, attempt {attempts}/{max_attempts}...")
raise Exception("Prodia job timed out")
elif self.image_gen == "hercai":
self.log("Using Hercai provider for image generation")
url = f"https://hercai.onrender.com/{self.image_model}/text2image?prompt={prompt}"
r = requests.get(url)
if r.status_code != 200:
raise Exception(f"Hercai API error: {r.text}")
parsed = r.json()
if "url" in parsed and parsed["url"]:
self.log("Image URL received from Hercai")
image_url = parsed["url"]
img_data = requests.get(image_url).content
with open(image_path, "wb") as f:
f.write(img_data)
self.images.append(image_path)
self.log(success(f"Image saved to: {image_path}"))
return image_path
else:
raise Exception("No image URL in Hercai response")
elif self.image_gen == "g4f":
self.log("Using G4F provider for image generation")
from g4f.client import Client
client = Client()
response = client.images.generate(
model=self.image_model,
prompt=prompt,
response_format="url"
)
if response and response.data and len(response.data) > 0:
image_url = response.data[0].url
image_response = requests.get(image_url)
if image_response.status_code == 200:
with open(image_path, "wb") as f:
f.write(image_response.content)
self.images.append(image_path)
self.log(success(f"Image saved to: {image_path}"))
return image_path
else:
raise Exception(f"Failed to download image from {image_url}")
else:
raise Exception("No image URL received from G4F")
elif self.image_gen == "segmind":
self.log("Using Segmind provider for image generation")
api_key = os.environ.get("SEGMIND_API_KEY", "")
if not api_key:
raise ValueError("Segmind API key is not set. Please provide a valid API key.")
headers = {
"x-api-key": api_key,
"Content-Type": "application/json"
}
response = requests.post(
"https://api.segmind.com/v1/sdxl-turbo",
json={
"prompt": prompt,
"negative_prompt": "blurry, low quality, distorted face, text, watermark",
"samples": 1,
"size": "1024x1024",
"guidance_scale": 1.0
},
headers=headers
)
if response.status_code == 200:
with open(image_path, "wb") as f:
f.write(response.content)
self.images.append(image_path)
self.log(success(f"Image saved to: {image_path}"))
return image_path
else:
raise Exception(f"Segmind request failed: {response.status_code} {response.text}")
elif self.image_gen == "pollinations":
self.log("Using Pollinations provider for image generation")
response = requests.get(f"https://image.pollinations.ai/prompt/{prompt}{random.randint(1,10000)}")
if response.status_code == 200:
self.log("Image received from Pollinations")
with open(image_path, "wb") as f:
f.write(response.content)
self.images.append(image_path)
self.log(success(f"Image saved to: {image_path}"))
return image_path
else:
raise Exception(f"Pollinations request failed with status code: {response.status_code}")
else:
# No fallback, raise an exception for unsupported image generator
error_msg = f"Unsupported image generator: {self.image_gen}"
self.log(error(error_msg))
raise ValueError(error_msg)
def generate_speech(self, text, output_format='mp3') -> str:
"""Generate speech from text using the selected TTS engine."""
self.progress(0.6, desc="Creating voiceover")
self.log("Generating speech from text")
# Clean text
text = re.sub(r'[^\w\s.?!,;:\'"-]', '', text)
self.log(f"Using TTS Engine: {self.tts_engine}, Voice: {self.tts_voice}")
# Always save to the generation folder when available
if hasattr(self, 'generation_folder') and os.path.exists(self.generation_folder):
audio_path = os.path.join(self.generation_folder, f"speech_{uuid.uuid4()}_{int(time.time())}.{output_format}")
else:
# Use STORAGE_DIR if no generation folder
audio_path = os.path.join(STORAGE_DIR, f"speech_{uuid.uuid4()}_{int(time.time())}.{output_format}")
if self.tts_engine == "elevenlabs":
self.log("Using ElevenLabs provider for speech generation")
elevenlabs_api_key = os.environ.get("ELEVENLABS_API_KEY", "")
if not elevenlabs_api_key:
raise ValueError("ElevenLabs API key is not set. Please provide a valid API key.")
headers = {
"Accept": "audio/mpeg",
"Content-Type": "application/json",
"xi-api-key": elevenlabs_api_key
}
payload = {
"text": text,
"model_id": "eleven_turbo_v2", # Using latest and most capable model
"voice_settings": {
"stability": 0.5,
"similarity_boost": 0.5,
"style": 0.0,
"use_speaker_boost": True
},
"output_format": "mp3_44100_128", # Higher quality audio (44.1kHz, 128kbps)
"optimize_streaming_latency": 0 # Optimize for quality over latency
}
# Map voice names to ElevenLabs voice IDs
voice_id_mapping = {
"Sarah": "21m00Tcm4TlvDq8ikWAM",
"Brian": "hxppwzoRmvxK7YkDrjhQ",
"Lily": "p7TAj7L6QVq1fE6XGyjR",
"Monika Sogam": "Fc3XhIu9tfgOPOsU1hMr",
"George": "o7lPjDgzlF8ZAeSpqmaN",
"River": "f0k5evLkhJxrIRJXQJvy",
"Matilda": "XrExE9yKIg1WjnnlVkGX",
"Will": "pvKWM1B1sNRNTlEYYAEZ",
"Jessica": "A5EAMYWMCSsLNL1wYxOv",
"default": "21m00Tcm4TlvDq8ikWAM" # Default to Sarah
}
# Get the voice ID from mapping or use the voice name as ID if not found
voice_id = voice_id_mapping.get(self.tts_voice, self.tts_voice)
self.log(f"Using ElevenLabs voice: {self.tts_voice} (ID: {voice_id})")
response = requests.post(
url=f"https://api.elevenlabs.io/v1/text-to-speech/{voice_id}",
json=payload,
headers=headers
)
if response.status_code == 200:
with open(audio_path, 'wb') as f:
f.write(response.content)
self.log(success(f"Speech generated successfully using ElevenLabs at {audio_path}"))
else:
try:
error_data = response.json()
error_message = error_data.get('detail', {}).get('message', response.text)
error_status = error_data.get('status', 'error')
raise Exception(f"ElevenLabs API error ({response.status_code}, {error_status}): {error_message}")
except ValueError:
# If JSON parsing fails, use the raw response
raise Exception(f"ElevenLabs API error ({response.status_code}): {response.text}")
elif self.tts_engine == "gtts":
self.log("Using Google TTS provider for speech generation")
from gtts import gTTS
tts = gTTS(text=text, lang=self.language[:2].lower(), slow=False)
tts.save(audio_path)
elif self.tts_engine == "openai":
self.log("Using OpenAI provider for speech generation")
openai_api_key = os.environ.get("OPENAI_API_KEY", "")
if not openai_api_key:
raise ValueError("OpenAI API key is not set. Please provide a valid API key.")
from openai import OpenAI
client = OpenAI(api_key=openai_api_key)
voice = self.tts_voice if self.tts_voice else "alloy"
response = client.audio.speech.create(
model="tts-1",
voice=voice,
input=text
)
response.stream_to_file(audio_path)
elif self.tts_engine == "edge":
self.log("Using Edge TTS provider for speech generation")
import edge_tts
import asyncio
voice = self.tts_voice if self.tts_voice else "en-US-AriaNeural"
async def generate():
communicate = edge_tts.Communicate(text, voice)
await communicate.save(audio_path)
asyncio.run(generate())
else:
# No fallback, raise an exception for unsupported TTS engine
error_msg = f"Unsupported TTS engine: {self.tts_engine}"
self.log(error(error_msg))
raise ValueError(error_msg)
self.log(success(f"Speech generated and saved to: {audio_path}"))
self.tts_path = audio_path
return audio_path
def generate_subtitles(self, audio_path: str) -> dict:
"""Generate subtitles from audio using AssemblyAI."""
# If subtitles are disabled, return empty data with settings
if not self.subtitles_enabled:
self.log("Subtitles are disabled, skipping generation")
return {
"wordlevel": [],
"linelevel": [],
"settings": {
"font": self.subtitle_font,
"fontsize": self.font_size,
"color": self.text_color,
"bg_color": self.highlight_color if self.highlighting_enabled else None,
"position": self.subtitle_position,
"highlighting_enabled": self.highlighting_enabled,
"subtitles_enabled": self.subtitles_enabled
}
}
self.log("Generating subtitles from audio")
try:
import assemblyai as aai
# Check if API key is set
aai_api_key = os.environ.get("ASSEMBLYAI_API_KEY", "")
if not aai_api_key:
raise ValueError("AssemblyAI API key is not set. Please provide a valid API key.")
aai.settings.api_key = aai_api_key
config = aai.TranscriptionConfig(speaker_labels=False, word_boost=[], format_text=True)
transcriber = aai.Transcriber(config=config)
self.log("Submitting audio for transcription")
transcript = transcriber.transcribe(audio_path)
if not transcript or not transcript.words:
raise ValueError("Transcription returned no words.")
# Process word-level information
wordlevel_info = []
for word in transcript.words:
word_data = {
"word": word.text.strip(),
"start": word.start / 1000.0, # Convert from ms to seconds
"end": word.end / 1000.0 # Convert from ms to seconds
}
wordlevel_info.append(word_data)
self.log(success(f"Transcription successful. Got {len(wordlevel_info)} words."))
# Define constants for subtitle generation
# Handle random font selection if configured
if self.subtitle_font == "random":
FONT = choose_random_font()
self.log(f"Using random font: {FONT}")
else:
FONT = self.subtitle_font
FONTSIZE = self.font_size
COLOR = self.text_color
BG_COLOR = self.highlight_color if self.highlighting_enabled else None
FRAME_SIZE = (1080, 1920) # Vertical video format
# Constants for line splitting
MAX_CHARS = 30 # Maximum characters per line for vertical video format
MAX_DURATION = 3.0 # Maximum duration for a single line
MAX_GAP = 1.5 # Split if nothing is spoken for this many seconds
# Split text into lines
subtitles = []
line = []
line_duration = 0
for idx, word_data in enumerate(wordlevel_info):
word = word_data["word"]
start = word_data["start"]
end = word_data["end"]
line.append(word_data)
line_duration += end - start
temp = " ".join(item["word"] for item in line)
new_line_chars = len(temp)
duration_exceeded = line_duration > MAX_DURATION
chars_exceeded = new_line_chars > MAX_CHARS
if idx > 0:
gap = word_data['start'] - wordlevel_info[idx-1]['end']
maxgap_exceeded = gap > MAX_GAP
else:
maxgap_exceeded = False
if duration_exceeded or chars_exceeded or maxgap_exceeded:
if line:
subtitle_line = {
"text": " ".join(item["word"] for item in line),
"start": line[0]["start"],
"end": line[-1]["end"],
"words": line
}
subtitles.append(subtitle_line)
line = []
line_duration = 0
# Add remaining words as last line
if line:
subtitle_line = {
"text": " ".join(item["word"] for item in line),
"start": line[0]["start"],
"end": line[-1]["end"],
"words": line
}
subtitles.append(subtitle_line)
self.log(success(f"Generated {len(subtitles)} subtitle lines"))
# Return the subtitle data and settings
return {
"wordlevel": wordlevel_info,
"linelevel": subtitles,
"settings": {
"font": FONT,
"fontsize": FONTSIZE,
"color": COLOR,
"bg_color": BG_COLOR,
"position": self.subtitle_position,
"highlighting_enabled": self.highlighting_enabled,
"subtitles_enabled": self.subtitles_enabled
}
}
except Exception as e:
error_msg = f"Error generating subtitles: {str(e)}"
self.log(error(error_msg))
raise Exception(error_msg)
def create_subtitle_clip(self, subtitle_data, frame_size):
"""Create subtitle clips for a line of text with word-level highlighting."""
# Early return if subtitles are disabled
if not subtitle_data.get("settings", {}).get("subtitles_enabled", True):
self.log("Subtitles are disabled, skipping subtitle clip creation")
return []
settings = subtitle_data["settings"]
font_name = settings["font"]
fontsize = settings["fontsize"]
color = settings["color"]
bg_color = settings["bg_color"]
highlighting_enabled = settings["highlighting_enabled"]
# Pre-load font and calculate color values once
try:
font_path = os.path.join(FONTS_DIR, f"{font_name}.ttf")
if os.path.exists(font_path):
pil_font = ImageFont.truetype(font_path, fontsize)
else:
self.log(warning(f"Font {font_name} not found, using default"))
pil_font = ImageFont.load_default()
except Exception as e:
self.log(warning(f"Error loading font: {str(e)}"))
pil_font = ImageFont.load_default()
# Parse colors once
if color.startswith('#'):
text_color_rgb = tuple(int(color.lstrip('#')[i:i+2], 16) for i in (0, 2, 4))
else:
text_color_rgb = (255, 255, 255) # Default white
if bg_color and bg_color.startswith('#'):
bg_color_rgb = tuple(int(bg_color.lstrip('#')[i:i+2], 16) for i in (0, 2, 4))
else:
bg_color_rgb = (0, 0, 255) # Default blue
# Optimize text clip creation - cache clips for reuse
clip_cache = {}
def create_text_clip(text, bg_color=None, cache_key=None):
# Use cache when possible for better performance
if cache_key and cache_key in clip_cache:
return clip_cache[cache_key]
try:
# Get text size
text_width, text_height = pil_font.getbbox(text)[2:4]
# Add padding
padding = 10
img_width = text_width + padding * 2
img_height = text_height + padding * 2
# Create image with background color or transparent
if bg_color:
img = Image.new('RGB', (img_width, img_height), color=bg_color_rgb)
else:
img = Image.new('RGBA', (img_width, img_height), color=(0, 0, 0, 0))
# Draw text
draw = ImageDraw.Draw(img)
draw.text((padding, padding), text, font=pil_font, fill=text_color_rgb)
# Convert to numpy array for MoviePy
img_array = np.array(img)
clip = ImageClip(img_array)
# Cache result for reuse
if cache_key:
clip_cache[cache_key] = (clip, img_width, img_height)
return clip, img_width, img_height
except Exception as e:
self.log(warning(f"Error creating text clip: {str(e)}"))
# Create a simple colored rectangle as fallback
img = Image.new('RGB', (100, 50), color=(100, 100, 100))
img_array = np.array(img)
clip = ImageClip(img_array)
return clip, 100, 50
subtitle_clips = []
# Calculate position constants once
if settings["position"] == "top":
y_buffer = frame_size[1] * 0.1 # 10% from top
elif settings["position"] == "middle":
y_buffer = frame_size[1] * 0.4 # 40% from top
else: # bottom
y_buffer = frame_size[1] * 0.7 # 70% from top
max_width = frame_size[0] * 0.8 # 80% of frame width
# Group words by timing to reduce number of clips (optimization)
word_groups = {}
# Process each line more efficiently by grouping
for line_idx, line in enumerate(subtitle_data["linelevel"]):
# Group words by start/end times to reduce clip count
line_text = line["text"]
line_start = line["start"]
line_end = line["end"]
line_duration = line_end - line_start
# First pass: calculate word dimensions and break text into lines
lines_data = [] # Store data for each line (words, positions)
current_line = []
current_x = 0
for word_data in line["words"]:
word = word_data["word"]
# Calculate dimensions without creating image yet
word_width = pil_font.getbbox(word)[2] + 20 # Add padding
word_height = pil_font.getbbox(word)[3] + 20
# Check if word fits on current line
if current_x + word_width > max_width and current_line:
# Complete current line
lines_data.append({
"words": current_line.copy(),
"total_width": current_x,
"height": max(w["height"] for w in current_line) if current_line else word_height
})
current_line = []
current_x = 0
# Add word to current line
word_info = {
"word": word,
"width": word_width,
"height": word_height,
"start": word_data["start"],
"end": word_data["end"]
}
current_line.append(word_info)
current_x += word_width
# Add the last line if needed
if current_line:
lines_data.append({
"words": current_line,
"total_width": current_x,
"height": max(w["height"] for w in current_line)
})
# Second pass: Create clip for each line (batch processing)
current_y = y_buffer
for line_data in lines_data:
# Calculate center position for entire line
line_width = line_data["total_width"]
x_center = (frame_size[0] - line_width) / 2
# Create text clip for complete line (non-highlighted base)
line_text = " ".join(w["word"] for w in line_data["words"])
cache_key = f"line_{line_idx}_{line_text}"
line_clip, measured_width, _ = create_text_clip(line_text, None, cache_key)
# Position the line in the center
line_clip = line_clip.set_position((x_center, current_y))
line_clip = line_clip.set_start(line["start"]).set_duration(line_duration)
subtitle_clips.append(line_clip)
# Add highlighted words if enabled (more efficiently)
if highlighting_enabled and bg_color:
current_x = x_center
# Group words with same timing to reduce clip count
timing_groups = {}
for word_info in line_data["words"]:
timing_key = f"{word_info['start']:.3f}_{word_info['end']:.3f}"
if timing_key not in timing_groups:
timing_groups[timing_key] = []
timing_groups[timing_key].append((word_info, current_x))
current_x += word_info["width"]
# Create one clip per timing group instead of per word
for timing_key, word_group in timing_groups.items():
start_time, end_time = map(float, timing_key.split('_'))
# If only one word in this timing, create single highlight
if len(word_group) == 1:
word_info, x_pos = word_group[0]
word = word_info["word"]
cache_key = f"word_{word}"
highlight_clip, _, _ = create_text_clip(word, bg_color, cache_key)
highlight_clip = highlight_clip.set_position((x_pos, current_y))
highlight_clip = highlight_clip.set_start(start_time).set_duration(end_time - start_time)
subtitle_clips.append(highlight_clip)
else:
# Multiple words with same timing - try to batch if adjacent
# (This is an optimization for words that appear together)
continue_batch = True
batch_start_idx = 0
while continue_batch and batch_start_idx < len(word_group):
# Start a new batch
batch = [word_group[batch_start_idx]]
batch_x = word_group[batch_start_idx][1]
current_batch_end = batch_start_idx
# Try to extend batch with adjacent words
for i in range(batch_start_idx + 1, len(word_group)):
prev_word, prev_x = word_group[i-1]
curr_word, curr_x = word_group[i]
# Check if words are adjacent
if abs(prev_x + prev_word["width"] - curr_x) < 5: # Small tolerance
batch.append(word_group[i])
current_batch_end = i
else:
break
# Create clip for this batch
if len(batch) > 1:
# Multiple adjacent words - create single highlight
batch_text = " ".join(info[0]["word"] for info in batch)
batch_width = batch[-1][1] + batch[-1][0]["width"] - batch[0][1]
cache_key = f"batch_{batch_text}"
highlight_clip, _, _ = create_text_clip(batch_text, bg_color, cache_key)
highlight_clip = highlight_clip.set_position((batch_x, current_y))
highlight_clip = highlight_clip.set_start(start_time).set_duration(end_time - start_time)
subtitle_clips.append(highlight_clip)
else:
# Single word in batch
word_info, x_pos = batch[0]
word = word_info["word"]
cache_key = f"word_{word}"
highlight_clip, _, _ = create_text_clip(word, bg_color, cache_key)
highlight_clip = highlight_clip.set_position((x_pos, current_y))
highlight_clip = highlight_clip.set_start(start_time).set_duration(end_time - start_time)
subtitle_clips.append(highlight_clip)
# Move to next batch
batch_start_idx = current_batch_end + 1
if batch_start_idx >= len(word_group):
continue_batch = False
# Move to next line
current_y += line_data["height"] + 10
# Limit the number of subtitle clips to avoid memory issues
if len(subtitle_clips) > 200:
self.log(warning(f"Too many subtitle clips ({len(subtitle_clips)}), limiting to 200 for performance"))
subtitle_clips = subtitle_clips[:200]
self.log(f"Created {len(subtitle_clips)} subtitle clips (optimized)")
return subtitle_clips
def combine(self) -> str:
"""Combine images, audio, and subtitles into a final video."""
self.progress(0.8, desc="Creating final video")
self.log("Combining images and audio into final video")
try:
# Use RAM for temporary files if possible
import tempfile
temp_dir = tempfile.mkdtemp()
# Always save to the generation folder when available
if hasattr(self, 'generation_folder') and os.path.exists(self.generation_folder):
output_path = os.path.join(self.generation_folder, f"output_{int(time.time())}.mp4")
else:
output_path = os.path.join(STORAGE_DIR, f"output_{int(time.time())}.mp4")
# Check for required files
if not self.images:
raise ValueError("No images available for video creation")
if not hasattr(self, 'tts_path') or not self.tts_path or not os.path.exists(self.tts_path):
raise ValueError("No TTS audio file available")
# Load audio
tts_clip = AudioFileClip(self.tts_path)
max_duration = tts_clip.duration
# Calculate duration for each image
num_images = len(self.images)
req_dur = max_duration / num_images
# Process each image ONCE to create base clips (optimization)
self.log("Processing images (optimized)")
processed_clips = []
for image_path in self.images:
if not os.path.exists(image_path):
self.log(warning(f"Image not found: {image_path}, skipping"))
continue
try:
# Load and process image once
clip = ImageClip(image_path)
# Use lower FPS for slideshow-style videos
clip = clip.set_fps(15)
# Handle aspect ratio (vertical video for shorts)
aspect_ratio = 9/16 # Standard vertical video ratio
if clip.w / clip.h < aspect_ratio:
# Image is too tall, crop height
clip = crop(
clip,
width=clip.w,
height=round(clip.w / aspect_ratio),
x_center=clip.w / 2,
y_center=clip.h / 2
)
else:
# Image is too wide, crop width
clip = crop(
clip,
width=round(aspect_ratio * clip.h),
height=clip.h,
x_center=clip.w / 2,
y_center=clip.h / 2
)
# Use a more efficient resolution (still good for mobile)
clip = clip.resize((720, 1280))
processed_clips.append(clip)
except Exception as e:
self.log(warning(f"Error processing image {image_path}: {str(e)}"))
if not processed_clips:
raise ValueError("No valid images could be processed")
# Create sequence using processed clips, repeated as needed
self.log(f"Creating video sequence from {len(processed_clips)} clips")
final_clips = []
tot_dur = 0
while tot_dur < max_duration:
for base_clip in processed_clips:
duration = min(req_dur, max_duration - tot_dur)
if duration <= 0:
break
# Reuse the pre-processed clip with new duration
duration_clip = base_clip.set_duration(duration)
final_clips.append(duration_clip)
tot_dur += duration
if tot_dur >= max_duration:
break
# Create video from sequence
self.log(f"Concatenating {len(final_clips)} clips")
final_clip = concatenate_videoclips(final_clips)
final_clip = final_clip.set_fps(15) # Lower FPS for slideshow-style
# Process audio
final_audio = tts_clip
# Add background music if available and enabled
if hasattr(self, 'enable_music') and self.enable_music and self.music_file != "none":
music_path = None
if self.music_file == "random":
music_path = choose_random_music()
elif os.path.exists(os.path.join(MUSIC_DIR, self.music_file)):
music_path = os.path.join(MUSIC_DIR, self.music_file)
if music_path and os.path.exists(music_path):
self.log(f"Adding background music: {music_path}")
try:
music_clip = AudioFileClip(music_path)
# Loop music if it's shorter than the video
if music_clip.duration < max_duration:
num_loops = int(np.ceil(max_duration / music_clip.duration))
music_clip = concatenate_audioclips([music_clip] * num_loops)
# Trim music if it's longer than the video
music_clip = music_clip.subclip(0, max_duration)
# Set music volume
music_volume = getattr(self, 'music_volume', 0.1)
music_clip = music_clip.volumex(music_volume)
# Combine with TTS audio
final_audio = CompositeAudioClip([tts_clip, music_clip])
except Exception as e:
self.log(warning(f"Error processing music: {str(e)}"))
# Set final audio
final_clip = final_clip.set_audio(final_audio)
# Add subtitles if enabled - process more efficiently
if self.subtitles_enabled and hasattr(self, 'subtitle_data'):
self.log("Adding subtitles (optimized)")
subtitle_clips = self.create_subtitle_clip(self.subtitle_data, (720, 1280)) # Match new resolution
if subtitle_clips:
final_clip = CompositeVideoClip([final_clip] + subtitle_clips)
# Write final video with optimized settings
self.log("Writing final video file (optimized encoding)")
final_clip.write_videofile(
output_path,
fps=15, # Lower FPS for slideshow-style
codec="libx264",
audio_codec="aac",
threads=8, # More threads for faster encoding
preset="ultrafast", # Fastest encoding preset
ffmpeg_params=["-crf", "28"] # Lower quality for speed
)
# Clean up temporary directory
import shutil
try:
shutil.rmtree(temp_dir, ignore_errors=True)
except Exception:
pass
self.log(success(f"Video saved to: {output_path}"))
return output_path
except Exception as e:
error_msg = f"Error combining video: {str(e)}"
self.log(error(error_msg))
raise Exception(error_msg)
def generate_video(self) -> dict:
"""Generate complete video with all components."""
try:
self.log("Starting video generation process")
# Create a unique folder with sequential numbering
folder_num = 1
# Check existing folders to find the latest number
if os.path.exists(STORAGE_DIR):
existing_folders = [d for d in os.listdir(STORAGE_DIR) if os.path.isdir(os.path.join(STORAGE_DIR, d))]
numbered_folders = []
for folder in existing_folders:
try:
# Extract folder number from format "N_UUID"
if "_" in folder:
num = int(folder.split("_")[0])
numbered_folders.append(num)
except (ValueError, IndexError):
continue
if numbered_folders:
folder_num = max(numbered_folders) + 1
folder_id = f"{folder_num}_{str(uuid.uuid4())}"
self.generation_folder = os.path.join(STORAGE_DIR, folder_id)
os.makedirs(self.generation_folder, exist_ok=True)
self.log(f"Created generation folder: {self.generation_folder}")
try:
# Step 1: Generate topic
self.log("Generating topic")
self.generate_topic()
# Step 2: Generate script
self.progress(0.1, desc="Creating script")
self.log("Generating script")
self.generate_script()
# Step 3: Generate metadata
self.progress(0.2, desc="Creating metadata")
self.log("Generating metadata")
self.generate_metadata()
# Step 4: Generate image prompts
self.progress(0.3, desc="Creating image prompts")
self.log("Generating image prompts")
self.generate_prompts()
# Step 5: Generate images
self.progress(0.4, desc="Generating images")
self.log("Generating images")
for i, prompt in enumerate(self.image_prompts, 1):
self.progress(0.4 + 0.2 * (i / len(self.image_prompts)),
desc=f"Generating image {i}/{len(self.image_prompts)}")
self.log(f"Generating image {i}/{len(self.image_prompts)}")
self.generate_image(prompt)
# Step 6: Generate speech
self.progress(0.6, desc="Creating speech")
self.log("Generating speech")
self.generate_speech(self.script)
# Step 7: Generate subtitles
self.progress(0.7, desc="Generating subtitles")
if self.subtitles_enabled and hasattr(self, 'tts_path') and os.path.exists(self.tts_path):
self.subtitle_data = self.generate_subtitles(self.tts_path)
# Save subtitles to generation folder
if self.subtitle_data:
try:
# Save word-level subtitles
if 'wordlevel' in self.subtitle_data:
word_subtitles_path = os.path.join(self.generation_folder, "word_subtitles.json")
with open(word_subtitles_path, 'w') as f:
json.dump(self.subtitle_data['wordlevel'], f, indent=2)
self.log(f"Saved word-level subtitles to: {word_subtitles_path}")
# Save line-level subtitles
if 'linelevel' in self.subtitle_data:
line_subtitles_path = os.path.join(self.generation_folder, "line_subtitles.json")
with open(line_subtitles_path, 'w') as f:
json.dump(self.subtitle_data['linelevel'], f, indent=2)
self.log(f"Saved line-level subtitles to: {line_subtitles_path}")
except Exception as e:
self.log(warning(f"Error saving subtitles to generation folder: {str(e)}"))
# Step 8: Save content.txt with all metadata and generation info
self.progress(0.75, desc="Saving generation data")
try:
content_path = os.path.join(self.generation_folder, "content.txt")
with open(content_path, 'w', encoding='utf-8') as f:
f.write(f"NICHE: {self.niche}\n\n")
f.write(f"LANGUAGE: {self.language}\n\n")
f.write(f"GENERATED TOPIC: {self.subject}\n\n")
f.write(f"GENERATED SCRIPT:\n{self.script}\n\n")
f.write(f"GENERATED PROMPTS:\n")
for i, prompt in enumerate(self.image_prompts, 1):
f.write(f"{i}. {prompt}\n")
f.write("\n")
f.write(f"GENERATED METADATA:\n")
for key, value in self.metadata.items():
f.write(f"{key}: {value}\n")
self.log(f"Saved content.txt to: {content_path}")
except Exception as e:
self.log(warning(f"Error saving content.txt: {str(e)}"))
# Step 9: Combine all elements into final video with optimized rendering
self.progress(0.8, desc="Creating final video")
self.log("Combining all elements into final video (optimized rendering)")
# Clear memory before video rendering
import gc
gc.collect()
path = self.combine()
self.progress(0.95, desc="Finalizing")
self.log(f"Video generation complete. Files saved in: {self.generation_folder}")
# Return the result
return {
'video_path': path,
'generation_folder': self.generation_folder,
'title': self.metadata['title'],
'description': self.metadata['description'],
'subject': self.subject,
'script': self.script,
'logs': self.logs
}
except Exception as e:
error_msg = f"Error during video generation step: {str(e)}"
self.log(error(error_msg))
# Try to clean up any resources
self.cleanup_resources()
raise Exception(error_msg)
except Exception as e:
error_msg = f"Error during video generation: {str(e)}"
self.log(error(error_msg))
raise Exception(error_msg)
def cleanup_resources(self):
"""Clean up any resources to prevent memory leaks."""
try:
# Force close any remaining ImageMagick processes
import psutil
for proc in psutil.process_iter():
try:
# Check if process name contains ImageMagick or ffmpeg
if 'magick' in proc.name().lower() or 'ffmpeg' in proc.name().lower():
proc.kill()
except (psutil.NoSuchProcess, psutil.AccessDenied, psutil.ZombieProcess):
pass
# Force garbage collection
import gc
gc.collect()
except Exception as e:
self.log(warning(f"Error during resource cleanup: {str(e)}"))
pass
# Data for dynamic dropdowns
def get_text_generator_models(generator):
"""Get available models for the selected text generator."""
models = {
"gemini": [
"gemini-2.0-flash",
"gemini-2.0-flash-lite",
"gemini-1.5-flash",
"gemini-1.5-flash-8b",
"gemini-1.5-pro"
],
"g4f": [
"gpt-4",
"gpt-4o",
"gpt-3.5-turbo",
"llama-3-70b-chat",
"claude-3-opus-20240229",
"claude-3-sonnet-20240229",
"claude-3-haiku-20240307"
],
"openai": [
"gpt-4o",
"gpt-4-turbo",
"gpt-3.5-turbo"
]
}
return models.get(generator, ["default"])
def get_image_generator_models(generator):
"""Get available models for the selected image generator."""
models = {
"prodia": [
"sdxl",
"realvisxl",
"juggernaut",
"dreamshaper",
"dalle"
],
"hercai": [
"v1",
"v2",
"v3",
"lexica"
],
"g4f": [
"flux",
"dall-e-3",
"dall-e-2",
"midjourney"
],
"segmind": [
"sdxl-turbo",
"realistic-vision",
"sd3"
],
"pollinations": [
"default"
]
}
return models.get(generator, ["default"])
def get_tts_voices(engine):
"""Get available voices for the selected TTS engine."""
voices = {
"elevenlabs": [
"Sarah", # Female, American accent
"Brian", # Male, British accent
"Lily", # Female, British accent
"Monika Sogam", # Female, Indian accent
"George", # Male, American accent
"River", # Female, American accent
"Matilda", # Female, British accent
"Will", # Male, American accent
"Jessica" # Female, American accent
],
"openai": [
"alloy",
"echo",
"fable",
"onyx",
"nova",
"shimmer"
],
"edge": [
"en-US-AriaNeural",
"en-US-GuyNeural",
"en-GB-SoniaNeural",
"en-AU-NatashaNeural"
],
"gtts": [
"en",
"es",
"fr",
"de",
"it",
"pt",
"ru",
"ja",
"zh",
"hi"
]
}
return voices.get(engine, ["default"])
# Create the Gradio interface
def create_interface():
with gr.Blocks(theme=gr.themes.Soft(primary_hue="indigo", radius_size="lg"), title="YouTube Shorts Generator") as demo:
with gr.Row():
gr.Markdown(
"""
# 📱 YouTube Shorts Generator
Generate engaging YouTube Shorts videos with AI. Just provide a niche and language to get started!
"""
)
with gr.Row(equal_height=True):
# Left panel: Content Settings
with gr.Column(scale=2, min_width=500):
with gr.Group():
gr.Markdown("### 📝 Content")
niche = gr.Textbox(
label="Niche/Topic",
placeholder="What's your video about?",
value="Historical Facts"
)
language = gr.Dropdown(
choices=["English", "Spanish", "French", "German", "Italian", "Portuguese",
"Russian", "Japanese", "Chinese", "Hindi"],
label="Language",
value="English"
)
# Generator Settings
with gr.Group():
gr.Markdown("### 🔧 Generator Settings")
with gr.Tabs():
with gr.TabItem("Text"):
text_gen = gr.Dropdown(
choices=["g4f", "gemini", "openai"],
label="Text Generator",
value="g4f"
)
text_model = gr.Dropdown(
choices=get_text_generator_models("g4f"),
label="Text Model",
value="gpt-4"
)
with gr.TabItem("Image"):
image_gen = gr.Dropdown(
choices=["g4f", "prodia", "hercai", "segmind", "pollinations"],
label="Image Generator",
value="g4f"
)
image_model = gr.Dropdown(
choices=get_image_generator_models("g4f"),
label="Image Model",
value="flux"
)
with gr.TabItem("Speech"):
tts_engine = gr.Dropdown(
choices=["edge", "elevenlabs", "gtts", "openai"],
label="Speech Generator",
value="edge"
)
tts_voice = gr.Dropdown(
choices=get_tts_voices("edge"),
label="Voice",
value="en-US-AriaNeural"
)
with gr.TabItem("Audio"):
enable_music = gr.Checkbox(label="Enable Background Music", value=True)
# Fix for music_file - Get available music and set proper default
music_choices = get_music_files()
default_music = "none" if "random" not in music_choices else "random"
music_file = gr.Dropdown(
choices=music_choices,
label="Background Music",
value=default_music,
interactive=True
)
music_volume = gr.Slider(
minimum=0.0,
maximum=1.0,
value=0.1,
step=0.05,
label="Background Music Volume"
)
with gr.TabItem("Subtitles"):
subtitles_enabled = gr.Checkbox(label="Enable Subtitles", value=True)
highlighting_enabled = gr.Checkbox(label="Enable Word Highlighting", value=True)
subtitle_font = gr.Dropdown(
choices=get_font_files(),
label="Font",
value="random"
)
with gr.Row():
font_size = gr.Slider(
minimum=40,
maximum=120,
value=80,
step=5,
label="Font Size"
)
subtitle_position = gr.Dropdown(
choices=["bottom", "middle", "top"],
label="Position",
value="bottom"
)
with gr.Row():
text_color = gr.ColorPicker(label="Text Color", value="#FFFFFF")
highlight_color = gr.ColorPicker(label="Highlight Color", value="#0000FF")
# Generate button
generate_btn = gr.Button("🎬 Generate Video", variant="primary", size="lg")
# Right panel: Output display
with gr.Column(scale=1, min_width=300):
with gr.Tabs():
with gr.TabItem("Video"):
# Larger video preview with proper mobile proportions
video_output = gr.Video(label="Generated Video", height=580, width=330)
with gr.TabItem("Metadata"):
title_output = gr.Textbox(label="Title", lines=2)
description_output = gr.Textbox(label="Description", lines=4)
script_output = gr.Textbox(label="Script", lines=8)
# API Keys section as a tab
with gr.TabItem("🔑 API Keys"):
gemini_api_key = gr.Textbox(
label="Gemini API Key",
type="password",
value=os.environ.get("GEMINI_API_KEY", "")
)
assemblyai_api_key = gr.Textbox(
label="AssemblyAI API Key",
type="password",
value=os.environ.get("ASSEMBLYAI_API_KEY", "")
)
elevenlabs_api_key = gr.Textbox(
label="ElevenLabs API Key",
type="password",
value=os.environ.get("ELEVENLABS_API_KEY", "")
)
segmind_api_key = gr.Textbox(
label="Segmind API Key",
type="password",
value=os.environ.get("SEGMIND_API_KEY", "")
)
openai_api_key = gr.Textbox(
label="OpenAI API Key",
type="password",
value=os.environ.get("OPENAI_API_KEY", "")
)
with gr.TabItem("Log"):
log_output = gr.Textbox(label="Process Log", lines=15, max_lines=100)
# Dynamic dropdown updates
def update_text_models(generator):
return gr.Dropdown(choices=get_text_generator_models(generator))
def update_image_models(generator):
return gr.Dropdown(choices=get_image_generator_models(generator))
def update_tts_voices(engine):
return gr.Dropdown(choices=get_tts_voices(engine))
# Connect the change events
text_gen.change(fn=update_text_models, inputs=text_gen, outputs=text_model)
image_gen.change(fn=update_image_models, inputs=image_gen, outputs=image_model)
tts_engine.change(fn=update_tts_voices, inputs=tts_engine, outputs=tts_voice)
# Main generation function
def generate_youtube_short(niche, language, text_gen, text_model, image_gen, image_model,
tts_engine, tts_voice, subtitles_enabled, highlighting_enabled,
subtitle_font, font_size, subtitle_position,
text_color, highlight_color, music_file,
enable_music, music_volume,
gemini_api_key, assemblyai_api_key,
elevenlabs_api_key, segmind_api_key, openai_api_key,
progress=gr.Progress()):
if not niche.strip():
return {
video_output: None,
title_output: "ERROR: Please enter a niche/topic",
description_output: "",
script_output: "",
log_output: "Error: Niche/Topic is required. Please enter a valid topic and try again."
}
# Create API keys dictionary
api_keys = {
'gemini': gemini_api_key,
'assemblyai': assemblyai_api_key,
'elevenlabs': elevenlabs_api_key,
'segmind': segmind_api_key,
'openai': openai_api_key
}
try:
# Initialize YouTube class
yt = YouTube(
niche=niche,
language=language,
text_gen=text_gen,
text_model=text_model,
image_gen=image_gen,
image_model=image_model,
tts_engine=tts_engine,
tts_voice=tts_voice,
subtitle_font=subtitle_font,
font_size=font_size,
text_color=text_color,
highlight_color=highlight_color,
subtitles_enabled=subtitles_enabled,
highlighting_enabled=highlighting_enabled,
subtitle_position=subtitle_position,
music_file=music_file,
enable_music=enable_music,
music_volume=music_volume,
api_keys=api_keys,
progress=progress
)
# Generate video
result = yt.generate_video()
# Check if video was successfully created
if not result or not result.get('video_path') or not os.path.exists(result.get('video_path', '')):
return {
video_output: None,
title_output: "ERROR: Video generation failed",
description_output: "",
script_output: "",
log_output: "\n".join(yt.logs)
}
return {
video_output: result['video_path'],
title_output: result['title'],
description_output: result['description'],
script_output: result['script'],
log_output: "\n".join(result['logs'])
}
except Exception as e:
import traceback
error_details = f"Error: {str(e)}\n\n{traceback.format_exc()}"
return {
video_output: None,
title_output: f"ERROR: {str(e)}",
description_output: "",
script_output: "",
log_output: error_details
}
# Connect the button click event
generate_btn.click(
fn=generate_youtube_short,
inputs=[
niche, language, text_gen, text_model, image_gen, image_model,
tts_engine, tts_voice, subtitles_enabled, highlighting_enabled,
subtitle_font, font_size, subtitle_position, text_color, highlight_color, music_file,
enable_music, music_volume, gemini_api_key, assemblyai_api_key, elevenlabs_api_key, segmind_api_key, openai_api_key
],
outputs=[video_output, title_output, description_output, script_output, log_output]
)
# Add examples
music_choices = get_music_files()
default_music = "none" if "random" not in music_choices else "random"
gr.Examples(
[
["Historical Facts", "English", "g4f", "gpt-4", "g4f", "flux", "edge", "en-US-AriaNeural", True, True, "default", 80, "bottom", "#FFFFFF", "#0000FF", default_music, True, 0.1],
["Cooking Tips", "English", "g4f", "gpt-4", "g4f", "flux", "edge", "en-US-AriaNeural", True, True, "default", 80, "bottom", "#FFFFFF", "#FF0000", default_music, True, 0.1],
["Technology News", "English", "g4f", "gpt-4", "g4f", "flux", "edge", "en-US-GuyNeural", True, True, "default", 80, "bottom", "#FFFFFF", "#00FF00", default_music, True, 0.1],
],
[niche, language, text_gen, text_model, image_gen, image_model, tts_engine, tts_voice,
subtitles_enabled, highlighting_enabled, subtitle_font, font_size,
subtitle_position, text_color, highlight_color, music_file, enable_music, music_volume],
label="Quick Start Templates"
)
return demo
# Create and launch the interface
if __name__ == "__main__":
# Create necessary directories
os.makedirs(STATIC_DIR, exist_ok=True)
os.makedirs(MUSIC_DIR, exist_ok=True)
os.makedirs(FONTS_DIR, exist_ok=True)
os.makedirs(STORAGE_DIR, exist_ok=True)
# Launch the app
demo = create_interface()
demo.launch()