File size: 3,341 Bytes
81000a9 2224634 f017bfe 2224634 81000a9 0b410bd 2224634 81000a9 2224634 81000a9 11a3ad8 2224634 f017bfe 2224634 f017bfe 2224634 11a3ad8 0b410bd 2224634 0b410bd 81000a9 2224634 11a3ad8 2224634 11a3ad8 2224634 81000a9 2224634 0b410bd 11a3ad8 0b410bd 81000a9 2224634 81000a9 0b410bd 2224634 11a3ad8 2224634 11a3ad8 2224634 0b410bd 81000a9 0b410bd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 |
import os
import gradio as gr
import yt_dlp
import whisper
from transformers import pipeline, MarianMTModel, MarianTokenizer
import torch
# Load Whisper model
whisper_model = whisper.load_model("small")
# Load summarizer
summarizer = pipeline("summarization", model="Falconsai/text_summarization")
# Load translation model (multilingual to English)
translation_tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-mul-en")
translation_model = MarianMTModel.from_pretrained("Helsinki-NLP/opus-mt-mul-en")
def download_audio(youtube_url):
output_file = "audio.webm"
ydl_opts = {
'format': 'bestaudio/best',
'outtmpl': output_file,
'quiet': True,
}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
ydl.download([youtube_url])
return output_file
def get_thumbnail(youtube_url):
ydl_opts = {'quiet': True}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
info = ydl.extract_info(youtube_url, download=False)
return info.get("thumbnail", "")
def translate_to_english(text):
chunks = [text[i:i+500] for i in range(0, len(text), 500)]
translated = []
for chunk in chunks:
inputs = translation_tokenizer(chunk, return_tensors="pt", truncation=True, max_length=512)
output = translation_model.generate(**inputs, max_length=512)
translated.append(translation_tokenizer.decode(output[0], skip_special_tokens=True))
return " ".join(translated)
def process_video(url):
audio_path = download_audio(url)
result = whisper_model.transcribe(audio_path)
transcription = result["text"]
translated_text = translate_to_english(transcription)
# Summarize
summary = summarizer(translated_text, max_length=130, min_length=30, do_sample=False)[0]["summary_text"]
# Get thumbnail
thumbnail_url = get_thumbnail(url)
return transcription, translated_text, summary, thumbnail_url
def download_summary(text):
filename = "summary.txt"
with open(filename, "w", encoding="utf-8") as f:
f.write(text)
return filename
# UI
with gr.Blocks(theme=gr.themes.Soft(), title="🎥 YouTube Video Summarizer with LLaMA") as demo:
gr.Markdown("## 🧠 Multilingual YouTube Summarizer")
gr.Markdown("Upload a video link and get the transcript, English translation, and summary.")
with gr.Row():
youtube_input = gr.Text(label="YouTube Video URL", placeholder="https://www.youtube.com/watch?v=...")
submit_btn = gr.Button("Transcribe & Summarize")
with gr.Row():
with gr.Column():
transcript_output = gr.Textbox(label="🔊 Original Transcript", lines=10)
translation_output = gr.Textbox(label="🌍 Translated to English", lines=10)
summary_output = gr.Textbox(label="🧾 Summary", lines=10)
download_btn = gr.Button("📥 Download Summary")
download_file = gr.File(label="Download Link")
video_thumb = gr.Image(label="🎞️ Video Thumbnail", width=256)
# Button actions
submit_btn.click(
fn=process_video,
inputs=[youtube_input],
outputs=[transcript_output, translation_output, summary_output, video_thumb]
)
download_btn.click(
fn=download_summary,
inputs=[summary_output],
outputs=[download_file]
)
demo.launch(share=True)
|